Article Text
PDF
Abstract
Urban or rural locality has been suggested to influence musculoskeletal health, with lower bone mineral density (BMD) and greater prevalence of fracture identified in urban residents. A computer-aided search of Medline, EMBASE, CINAHL and PsychINFO, January 1966 to November 2007 was conducted to identify studies investigating the relationship between urban or rural locality and the occurrence of hip fracture. The methodological quality of studies was assessed, and a best-evidence synthesis was used to summarise the results. Fourteen cohort studies and one case-control study were identified for inclusion in this review, indicating a lack of literature in the field. Best-evidence analysis identified moderate evidence for residents of rural regions to have lower risk of hip fracture compared to urban residents. Examining principal mechanisms for the observed relationship between urban/rural locality and hip fracture, such as factors at the person or area level, may help to identify modifiable risk factors and inform appropriate prevention strategies.
- Urban or rural locality
- hip fracture
- systematic review
- fractures
- osteoporosis
- urban health
Statistics from Altmetric.com
Introduction
Hip fractures have significant attributable morbidity and mortality, and are a major public health burden.1 A relationship exists between hip fracture and a significant decrease in quality of life, an increased likelihood of death for the aged and a greater demand on health infrastructure.2–6 Estimated financial costs associated with hip fracture have been widely documented, and are expected to increase as the population ages.7 8
Much attention has been directed towards urban or rural locality as being influential upon general health9–11 and an urban location has been associated with lower bone mineral density (BMD) and greater fracture incidence. The relationship between risk of hip fracture and living in an urban or rural region has been examined in a number of countries, and speculation is wide regarding why differences in fracture rates between regions have been observed. Causal factors have been suggested to include person-level factors such as occupation, lifestyle choices or comorbidities, and area-level factors including environmental hazards. The conflicting views regarding causal factors are compounded by some inconsistencies in the available literature. A cohort study by Sanders et al12 identified lower fracture rates for rural populations, whereas a study by Larsson et al13 identified higher fracture rates for rural populations. Unless mechanisms underlying the relationship between urban/rural locality and hip fracture are better understood, effective intervention to reduce the disproportionate burden of fracture between regions is limited. The aim of this study was to conduct a critical review of the available evidence, in order to inform future health policy and reduce the burden on healthcare services due to fracture incidence.4 14 15
Materials and methods
Criteria for considering studies
Epidemiological cohort, case-control or cross-sectional studies in which individuals of advanced age (>50 years) had suffered a fracture of the hip and in which an urban or rural location were the exposure of interest were included. Hip regions included the neck or head of femur, trochanteric (intertrochanteric and pertrochanteric), and cervical regions (subcapital, transcervical or basicervical). Analyses addressing semi-rurality, or other variations of urban or rural locality that did not use administrative classification and/or population density to define regions were excluded from this review, as these distinctions between regions were non-definitive.
Search strategy for identification of studies
A computerised search strategy was performed using Medline, EMBASE, CINAHL and PsychINFO between January 1966 and 30 November 2007 to identify relevant studies. Relevant truncation was used for each database, and the following key MeSH terms; hip fracture, and urban rural difference.
References of articles that fulfilled the eligibility criteria were hand-searched, and citations of those published in the field of interest were tracked. Two reviewers (SLB and AEW) confirmed the selection of articles.
Methodological quality
To assess the methodological quality of the included studies, two reviewers (SLB and FH) independently scored them using a modified scoring system as previously applied by Lievense et al16 17 (table 1), and used in a systematic review of the relationship between fracture and socioeconomic status.18 The preferred design was a cohort study, followed by a case-control study and then a cross-sectional study design. Each of the 14 items was scored as positive (1) or negative (0), and the methodological strength of the study design was evident in the maximum possible score. Several of the criteria were not applicable to certain study types, thus the maximum score for each study was based only on the items applicable to that particular type of study design. Where the reviewers disagreed and could not achieve consensus, a third reviewer (AEW) gave a final judgement. High quality was defined as achieving a score above the mean of all quality scores.
Criteria list for assessment of study quality, adapted from Lievense et al17
Data analysis
As there was much heterogeneity in the available studies, no meta-analysis was performed but a ‘best-evidence synthesis’ was undertaken (table 2). This synthesis consisted of five levels of evidence, ranging from ‘no’ to ‘strong’ evidence and primarily reflected the study designs used.17
Criteria list for determining the level of evidence for best evidence synthesis, adapted from Lievense et al17
Results
The computer-assisted search yielded 1619 articles, of which 38 articles were potentially eligible for inclusion. Twenty-three articles did not meet the criteria as they had analysed geographic regions not defined as urban/rural, or analysed only one of the two (n=13)19–31; examined ethnicity or international variations (n=4)32–35; were not a community-based study or compared institutionalised older people with community-based older people (n=4)36–39; or analysed anatomical sites other than hips and/or did not separate hip fractures from other anatomical sites during analyses (n=3).40–42 Thus, 15 articles met the selection criteria. The reference lists of these articles were hand-searched; however, no further studies were identified as eligible through this strategy. Citations of these 15 studies were electronically tracked; however, no further studies were identified. This left the final number of studies to be included at fifteen.
The characteristics of the included studies (n=15) are presented in table 3. Only five of the studies were published during the last decade.43 51–54 Fourteen studies were cohort studies12 13 43–54 and the remaining study was a case-control design.55 Nine of the studies were performed in Scandinavian countries13 43–47 49 50 54 and the other six examined populations from America,48 51 Australia,12 52 Switzerland53 and Turkey.55 The age of participants across all studies ranged from 50 to 96 years. Only three of the 13 studies provided mean (SD) data for participant ages, indicating the range to be 67.1 (7.9) to 83.4 (9.7).45 53 55 The gender breakdown in the age group of interest was not provided for two studies,45 51 although of the remaining studies 50.2–80.0% of subjects were women. Four of the studies were designed to compare fracture rates across multiple time periods.13 44 46 54
Study characteristics of the included manuscripts, including country, time period, population data, case numbers, age and method of fracture identification
Urban and rural localities were defined in various ways across all the identified studies. Whereas two studies defined rural localities based solely on a high level of agricultural, forestry or fishing activity,44 54 the majority defined rural localities by low population density,45–47 50 52 53 55 and six studies applied both criteria.12 13 43 48 49 51 Table 3 outlines definitions and cut points applied by different studies to identify urban and rural regions. Population density is the primary index used in social science research by Scandinavian countries to distinguish between urban and rural regions.49 Various and/or multiple methods were used to identify hip fractures, including reviewing hospital admission or discharge records for International Classification of Disease (ICD) codes (n=9), radiology reports (n=8), surgery reports and/or individual patient records (n=3). Eight of the studies standardised to the overall age structure of the country of interest,12 44–46 48 51–53 three studies standardised to the separate age structure for urban/rural regions,43 47 49 one study standardised to the urban regions alone50 and the remaining two studies standardised to the country and urban/rural regions.13 54
The two reviewers (SLB and FH) scored a potential of 181 criteria within the included studies, and independently agreed on 155 items (85.6%). The discrepancies were resolved in a single consensus meeting. The quality assessment results of the studies are presented in table 4, in order of their methodological quality, subdivided into the different types of study designs. The scores ranged from 46%55 to 83%12 of the maximum obtainable score for each of the study designs. The mean of all quality scores was 63% (SD 9%). Using the mean quality score as the cut-off point, six of the fifteen studies were above this score, and thus determined to be high quality.12 48–52
Quality scores of studies in order of their methodological quality, and subdivided into the different types of study designs
Table 5 presents the results of the reviewed studies, ordered by study design and year of publication. Results are presented as annual incident rate per 10 000 person years (py), or OR or RR, with CIs and p values presented where available.
Summary of hip fracture incidence estimates and quality assessment results grouped by study design and presented by year of publication
Four of the studies13 44 46 54 had examined multiple time periods, therefore the 15 studies presented 22 different analyses. Of these, 15 gender-stratified analyses identified a reduced risk of hip fracture for men and women residing in rural locations.12 44–46 48 50 52–54 Similarly, a further three sex-adjusted analyses identified a reduced risk for individuals residing in rural locations.13 43 Two of the 22 analyses identified that residents of rural locations experienced an increased risk of hip fracture compared with residents of urban locations, although did not identify the degree of significance.13 55 Two analyses identified no difference in fracture rates between urban and rural regions.49 51 No differences in fracture rates were identified in relation to whether urban and rural regions were defined by administrative classifications, defined by the level of agricultural activities, and/or population density.
The most common methodological shortcoming was the difficulty in ascertaining from each study whether researchers were blinded to the exposure of an urban/rural locality when extracting fracture data. Other shortcomings were the retrospective design of the majority of reviewed studies when compared to prospective studies, which are stronger methodologically, and that four of the reviewed studies did not present data relating to 95% CIs or p values.13 46 51 55
Overall, the data represented much methodological heterogeneity, rendering comparisons between the studies difficult. As an alternative to meta-analysis, a best-evidence synthesis was performed to assess the strength of the association between findings, as has previously been published.16 17 56 High quality studies, as determined by methodological assessment quality score above the mean of 63%, were included in the best-evidence synthesis. The difficulty of statistically pooling the extracted data was confirmed by the heterogeneity that remained in the six studies determined as high quality.
All six studies determined as high quality studies were of cohort design.12 48–52 The age and gender adjusted studies by Kaastad et al,50 Cooley et al,52 Sanders et al12 and Madhok et al48 each identified significantly lower point estimates of fracture for individuals residing in rural compared with urban regions (p<0.05). The sex-adjusted analyses by Luthje et al49 and Melton et al51also identified lower point estimates; however, no significance was identified between urban or rural regions. The high quality studies represented populations from Australia,12 52 America,48 51 Finland49 and Norway,50 and suggest that residents of rural regions within these countries are less likely to experience hip fracture in comparison to their urban counterparts, independent of gender.
Discussion
Twelve of the 15 studies provided consistent evidence that residents of rural regions had lower incidence of hip fracture than residents of urban regions.12 43–50 52–54 In contrast, only two of the reviewed studies identified no difference between urban or rural regions,51 55 and one of the reviewed studies identified higher fracture incidence for rural residents.13 When applying a best-evidence analysis to the available literature, only six studies were identified of high quality; all but one of these provided consistent evidence of a lower incidence of fractures in rural regions compared with urban. Thus, it can be concluded from the available literature that a moderate level of evidence exists for a statistically significant lower point estimate of fracture incidence in rural residents compared with urban residents, as identified across multiple high quality cohort studies.
Causal factors of fracture rates may include person-level factors such as occupation,57–59 physical activity,12 41 43 44 greater serum vitamin D due to increased sunlight exposure which influences BMD,60–63 comorbidities including previous fracture64–66 or factors related to relocation.12 48 67 Equally, area-level characteristics associated with the locality of residence may influence fracture rates, including environmental hazards,68 wars or other forms of civil unrest,49 or influences upon BMD through pollution or climatic extremes,51 52 55 or the fluoridisation of water supplies.44 49 69 Each factor has the potential to influence the observed relationship between hip fracture and an urban or rural locality; however, interactions between person-level and area-level factors may influence fracture rates more than one factor alone.
A physically active lifestyle increases muscle strength and postural stability, both of which have been observed in rural residents compared with urban residents.25 44 70 Occupational physical activity influences muscle strength and neuromuscular function,58 71 and has also been associated with reduced fracture risk.57–59 It is possible that greater levels of physical activity are undertaken in rural regions compared to urban regions, most likely influenced by differences in living environments and the types of occupations available. However, although it is conceivable that agricultural occupations may be more physically demanding than urban-based occupations,41 43 44 the differences in levels of physical activity associated with all occupations are likely to be gradually reducing due to increased mechanisation. It is feasible that the definition, and practices, of agricultural occupations may also differ between countries. This is especially likely between non-Scandinavian and Scandinavian countries, and between non-Western and Western countries. Thus, assumptions of traditional distinctions of the urban/rural distinctions across countries should be made with caution.
Other lifestyle choices that may influence fracture rates include dietary or supplemental calcium, smoking and/or alcohol,72 and the use of medications including hormone therapy, oral corticosteroids or anti-resorptive medications.73–76 Although data are few regarding the relationship between these lifestyle factors and an urban or rural locality, it has been suggested that patterns of these exposures may vary depending upon the characteristics of a region.40 41
Comorbidities have been shown to increase the likelihood of fracture.65 It has been shown that, in comparison to residents of rural regions, urban residents have greater levels of comorbidities,48 predisposing them to more fragile bones and/or an increased likelihood of falls. The predominance of comorbidities in urban residents may be a result of ageing or declining health and independence influencing a move to cities for increased accessibility to healthcare services.12 48 67 Although the present study was unable to address this issue directly, some studies have attempted to examine a dose–response relationship between the length of time residing in urban or rural regions and an increased risk of fracture.41 49 53 Younger adults have also been shown to migrate towards urban regions to access increase increased opportunities for employment and socialisation.77 This age-related shift may confound analyses that do not account for the length of time residing in one region compared with another.
Environmental hazards, such as risks of being in crowded areas or needing to manoeuvre other hazardous obstacles may increase the risk of fracture within urban regions. Other environmental factors may influence fracture rates via mediating BMD, including the natural or imposed fluoridating of water supplies,44 49 69 or lifestyles associated with postwar or civil unrest.49
Although the available literature presented rather moderate evidence for lower fracture risk in rural regions, one study showed a greater incidence of hip fracture in rural regions.55 These findings are in the context of Turkey being a relatively disadvantaged country in 1988–89, with an influential factor for fracture incidence being limited occupational safety laws. With greater physical labour in rural regions, compared to urban regions, an increased risk of high trauma fractures related to occupation may have skewed the findings of fracture incidence in urban/rural regions. When this study restricted analysis to only low energy fractures, no differences were identified between urban and rural regions. Although low fracture incidence in Turkey was identified within the Mediterranean Osteoporosis Study (MEDOS), a healthy worker effect may have influenced these data.
The present methodology has been adapted from previous applied systematic reviews;16 17 however, several limitations are recognised. First, although much effort was invested in the development of the search strategy, relevant articles may have been missed due to the use of alternate key words, and/or the article not being indexed. Articles written in languages other than English were excluded from this review.
Due to limited available literature, there are further limitations of the findings of the present study. Although most of the identified studies were of cohort design, and as such, methodologically robust, a statistical comparison of the studies was inappropriate due to considerable heterogeneity, including populations, method of fracture identification, and definition of urban and rural locality. Nevertheless, the results were consistent in their direction, with all but one study showing a reduction in hip fracture rate in rural compared to urban dwellers. Many studies used radiological reports and hospital records to identify fracture, which is likely to influence an under-reporting of fractures, and also increases the likelihood of human error in data entry. Although this may be more marked in a rural locality where access to radiography may be limited for distal extremity fractures,52 in the case of hip fractures admission to hospital is likely to be required in both settings, so this limitation is not likely to play a significant role. Nine of the 15 reviewed studies were from Nordic countries, and due to these populations being relatively homogenous with some of the highest fracture incidences globally, this may represent a bias in comparison to studies performed in other countries. Despite this, the direction of effect was similar to that seen in the other studies.
This summary of the current evidence regarding hip fracture and urban or rural locality shows compelling data to suggest the occurrence of hip fractures are more likely to be experienced by urban residents than their rurally located counterparts. To more fully examine this field of enquiry, further population-based cohort studies are required. Identifying definitive reasons for lower fracture incidence in rural areas was beyond the scope of the present study; however, examining patterns in risk factors for low BMD across urban and rural regions might assist in the identification of modifiable risk factors attributable to differences in fracture risk. Elucidation of the variance between countries in social and cultural factors, potentially associated with fracture risk, is required in order to effectively intervene to reduce health inequities between urban and rural populations.
What is already known on this topic
Hip fractures significantly decrease quality of life, and produce a greater demand on health infrastructure.
The relationship between risk of hip fracture and living in an urban or rural locality has been examined in a number of different countries, however there are inconsistencies in the available literature.
What this study adds
This study synthesises studies examining the association between urban and rural locality and hip fractures.
This study shows compelling data that hip fractures are more likely to be experienced by urban residents than rural residents.
It provides a critical review of the available evidence to inform future health policy.
References
Footnotes
Funding Ms Brennan is supported by NHMRC PhD scholarship (519404). Associate Professor Pasco and Dr Wluka are the recipients of NHMRC Project Grant (436665). Dr Hanna, Dr Urquhart and Dr Wluka are the recipients of NHMRC Public Health and Clinical Fellowships (418961, 284402 and 317840).
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.