Low back pain risk factors in a large rural Australian Aboriginal community. An opportunity for managing co-morbidities?

A cross-sectional self-report survey (Kempsey survey) was conducted to determine the extent of risk factors (Table 1) and their association with LBP in the study community.

Participating community members completed a consent form that explained the purpose of the survey. Ethics approval was obtained from the Durri Aboriginal Corporation Medical Service (ACMS) Board of Directors and the Human Research Ethics Committee of the University of Newcastle.

The Durri Community of Kempsey, NSW, Australia, comprises one of Australia's largest rural Aboriginal communities. The Durri (ACMS) is at the forefront of providing culturally appropriate care, largely via its Aboriginal Health Workers (AHWs). Durri ACMS aims to:

'make primary health care and education accessible to all members of the community in a culturally appropriate and spiritually sensitive manner, endeavouring to improve not only the health status but also the well-being of the Durri Aboriginal community' [13].

Discussions with a cross section of community members led to the formation of a Community Advisory Group (CAG) (which included representatives from the Durri ACMS, Booroongen Djugun Aboriginal Health Worker College, Hands On Health Australia and the University of Newcastle). The CAG aimed to advise on the development and implementation of the musculo-skeletal prevalence study [14]. Aboriginal Health Workers were chosen as the study agents because they are recognised as essential in providing culturally appropriate and effective health-care for their communities [15‐22].

Community consultation occurred throughout the study. This process involved regular discussions with key-informants from the community including AHWs, elders and health professionals. The community was informed of developments via information sheets and the publication of a summary report during the process and at the completion of the study.

Our goal was to select a representative cross-sectional sample of the local Aboriginal community of sufficient size to generalise our major findings to the whole local community (population 550). A random sampling procedure stratifying for age and sex was used to derive a representative sample of the local community. The sample size was generated using Epi-Info 6 [23]. With a population size of 550, the expected frequency of the main variable of interest (low back pain) was estimated at 50%. The value chosen as the farthest acceptable from the real population was 44%. Using these values and a 95% confidence interval, the ideal random sample size calculated was 180. However, we expected that logistically this was unlikely to be achieved, as many of the sample selected were likely to be uncontactable given the transient nature of community residents [24]. Accordingly, where randomly selected community members were unable to participate, they were replaced using a convenience sampling approach to achieve the required sample size. Although this strategy was not ideal, all attempts were made to attain a representative sample. Participants within the community were selected from persons aged 15-years or older who had been previously identified as Aboriginal (according to the definition of Aboriginal adopted by the Department of Aboriginal Affairs Constitutional Section) [25]. These participants were recruited by distributing letters inviting them to contact the assisting AHWs at the ACMS. If no response was received within a week, an attempt to contact the person via telephone was made by the assisting AHW.

The Kempsey survey included a screening survey administered by Aboriginal Health Workers immediately followed by a clinical conducted by chiropractors blinded to the findings of the screening survey.

Those who consented to participate were asked to attend the Durri ACMS. If participants found transport to the ACMS difficult, either the research team (including the researcher, the AHW and volunteer chiropractors/chiropractic students) would travel to the participants' homes, or the assisting AHW would arrange for the Durri ACMS bus to provide transportation at no charge.

Participants completed a screening survey previously found to be culturally acceptable and sensitive in measuring musculo-skeletal conditions and associated risk factors in this community. The survey achieved satisfactory measurement agreement (Kappa scores) when compared to a clinical assessment performed by chiropractors (a proxy "Gold Standard") [22]. Although some authors argue that a 'Gold standard' does not exist in many areas of musculo-skeletal practice [26], standard clinical assessments performed by musculo-skeletal health professionals provide the best available tools for measuring painful and limited ranges of motion and a provisional diagnosis [27]. The purpose of the screening survey was to identify those who had experienced a musculo-skeletal condition including ache, pain or discomfort. The questionnaire also assessed self-reported limitations to Activities of Daily Living (ADL) imposed by pain.

Participants screened by the AHW-administered survey subsequently underwent a clinical examination conducted by four chiropractors previously trained and assessed in standard, clinical assessment procedures according to a procedural manual which outlined the cultural considerations and logistical processes required by researchers. The content of the procedural manual was revised in a two-hour workshop for participating researchers to clarify and standardise study requirements. The exam was based on accepted clinical parameters for conducting musculoskeletal conditions and included the domains of assessment used by teaching institutions [28]. Thus attempts were made to fulfil content and face validity.

Participants attended a clinical assessment immediately following the screening survey to confirm the presence of musculo-skeletal conditions [22]. Chiropractors and 5th year chiropractic students performed a follow-up clinical assessment (based on clinical assessment parameters used in 1999 at the School of Chiropractic, RMIT University, Victoria, Australia) [28] to validate the findings reported in the screening questionnaire.

A positive pain finding in the clinical assessment was derived by practitioner-based examination, including the patient's history of involved site(s) followed by standard orthopaedic and range of motion tests to localise sites of pain and restricted movement. A negative pain finding was indicated by the absence of reported pain and/or restricted orthopaedic and range of motion findings as examined by the practitioner. Trivial LBP was differentiated from important LBP using a Likert scale. High levels of pain were interpreted as those ranging between 6–10 on a Likert scale of 0–10. Only those reporting "High" levels of pain were analysed in this study. Further questions related to any musculo-skeletal condition(s) experienced in the last seven days. In particular, probable causes of symptoms, past history, initial episode(s) of symptoms, duration of symptom(s), 'average' severity of symptoms and any associated limitation of daily activities. Also examined were, social routine and work activities, the type of treatment received and any barriers to receiving treatment were sought.

In the history component of the clinical assessment, chiropractors once again questioned participants about the presence of musculo-skeletal risk factors (according to the criteria reported in Table 1). Risk factor data were derived in the history component of the clinical assessment by asking questions from a list of modifiable occupational and lifestyle factors. Results for LBP as measured in the clinical assessment were used in the analysis. Clinical findings requiring follow-up treatment, management or referral was also identified.

Health workers using a laptop computer entered data on-site into a specifically designed, Microsoft Access database.

The questionnaire results were compared to the data from the clinical examination and published in a previous study (Table 2). Eighty-three percent of all participants reporting LBP in the screening survey also tested positive for LBP via the clinical assessment. Sensitivity of the screening survey for LBP was 0.83, specificity 0.63 and Kappa 0.46. Thus the screening survey achieved an adequate level of agreement with the clinical assessment [29].

The main variables of interest from the survey and clinical assessment were:

• Demographic and other sample characteristics-age, sex, number of children, occupation, weight, and Body Mass Index (BMI).

• Prevalence of LBP (within the last seven days, according to self report).

• Pain levels were recorded using a Likert scale where a score of 0 corresponded to no pain and 10 to severe pain.

• Duration of LBP was categorised as less than/equal to or more than seven weeks.

• Disability levels were recorded using a Likert scale where a score of 0 corresponded to no disability and 10 to severe disability. Disability was defined as "how much the condition (ache, pain or discomfort) had affected the participants ability to carry out daily activities (e.g., housework, washing, dressing, lifting, walking, driving, climbing stairs, getting in and out of bed or a chair, sleeping, working, social activities and sport)".

• Self-reported modifiable risk factors as described in Table 1 (according to a standardised clinical history).

• Other musculo-skeletal conditions.

Frequencies and confidence intervals were reported for characteristics of the sample, prevalence of LBP and reported risk factors for low back pain. Chi-square analyses were performed to test for factors associated with low back pain. Given the number of variables, only significant associations were reported.

The study was conducted between January 2001 and July 2002. The sample comprised 189 Indigenous people: 80 were selected randomly and the remainder were convenience sampled as described above.

Even though a trend was evident, no statistical association between LBP and the lifestyle factors detailed above. However, more participants reporting high levels of LBP were overweight or obese and obesity was statistically associated with self-reported strain causing reported LBP (χ² = 9.02, df = 2 10, p = 0.01). While sporting injuries were not statistically associated with report of LBP in particular, participants reporting sporting injuries experienced between two and four musculo-skeletal conditions (χ² = 7.90, df = 2, p = 0.02).