Chapter 3 Methodology
3.10 Data collection instruments
A standard data collection form (Appendix 3.D) was administered by a trained interviewer and a standardized clinical examination and relevant laboratory and radiological investigations were performed.
Information recorded included demographic data, anthropometric measures, risk factors for osteoporosis, functional assessment, full physical examination findings, results of haematological, biochemical and radiological investigations.
3.10.1 Demographic data
The following information was obtained for both hip fracture and control subjects.
1. Age
2. Ethnic group 3. Gender
4. Physical address was recorded as the usual address of the subjects.
Subjects outside the eThekwini boundary were excluded from the study.
5. Housing type : categorized as formal, informal, traditional or hostel.
6. Employment: categorized as pensioner, unemployed or self-employed.
Permanent South African citizens aged 60 years and older are eligible to receive an older person grant from the government and free health care [471].
7. Educational level: categorized as never schooled, less than 5 years of education, completed secondary level or Standard 10 or tertiary education.
3.10.2 Anthropometric measures
1. Weight: Subjects were weighed on a balanced beam scale, wearing the minimum of clothes, to the nearest kilogram (kg)
2. Height : was measured using a stadiometer with a sliding headpiece to increase the accuracy of the reading and taken to the nearest centimetre (cm).
Weight and height in hospital were recorded in 136 subjects only. Sixty four hip fracture subjects were unable to stand independently and to be accurately weighed by the time of discharge.
3. Body Mass index (BMI) was calculated using the standard formula BMI = weight (kg) / height (m2) and expressed as kg/m2 and according to the WHO categories for adults aged 25 years and older (Table 3.3) [472]
Adults: >25 year
Adapted from WHO 1995 [472]
3.10.3 Risk factors for osteoporosis
Although a number of validated tools for the assessment of osteoporotic fracture risk exist, few have been used in clinical practice in SA. The WHO FRAX® has been validated by the IOF and is based on studies from NA, Europe, Australia and Asia
Table 3.3 WHO normal reference values for BMI in adults
Category BMI range (kg/m2)
Underweight <19.0
Normal (healthy weight) 19.1 - 24.9
Overweight > 25 - 29.9
Obese > 30
[31]. The osteoporosis risk factor assessment in this study was based on traditional risk factors for osteoporosis including those identified by the FRAX® [78, 79]. The FRAX® tool has been validated in 12 independent studies and found to be valid and reliable.
Therefore the factors used in predicting fracture risk in this study included:
Age, sex, weight and height
o The actual weight and height were recorded.
o Subjects were categorized as being </> 57 kg as per the FRAX® tool, however only mean weight was used in the risk assessment analysis.
Gynaecological history o age of menarche o age of menopause o use of HRT
Age at which commenced use, duration and side effects of HRT o parity
History of childhood fractures
Previous fragility fractures after the age of 40 years
o defined by site, date of occurrence and treatment received o screened or treated for osteoporosis
Prior vertebral fractures (self-reported)
o defined site, date, treatment received o screened or treated for osteoporosis o kyphosis(self-reported)
o Family history o Osteoporosis
o Maternal history of falls o Maternal history of fractures o Smoking history
o The WHO Monitoring Trends and Determinants in Cardiovascular disease scale (MONICA scale), an internationally validated questionnaire was used to classify subjects as present or past smokers. The average number of cigarettes smoked was determined.
The age of smoking, exposure to passive smoking and use of a pipe was also documented [473].
Alcohol use
o Subjects were asked to report the number and type of alcoholic drinks consumed in the past week using a self-report scale, validated in The Danish Health and Morbidity Survey [474]. This scale was used as recall periods of greater than one week duration are associated with a significant decrease in the report of alcohol usage and are less accurate.
Calcium intake
o The IOF Calcium Intake Diary [475] was used to record and calculate the dietary calcium intake based on the UK nutrient databank information. While calcium intake has been reviewed in the SA population, several different methods were used to quantify calcium intake in the different ethnic groups. Despite the IOF calcium intake
tool being freely available this has not been used in previous studies in SA.
Lifestyle factors
o Illicit drug use, caffeine intake and sunlight exposure (minutes/day) was recorded.
Activity level (self-reported)
o Subjects were asked to self-categorize their activity level as extremely active, moderately active, mildly active or sedentary based on their daily activities.
Current drug use and previous use of drugs associated with bone toxicity, GC, anti-epileptics, heparin, lithium, antidepressants, was recorded.
Presence of known causes of secondary osteoporosis including rheumatoid arthritis, type 1 DM, osteogenesis imperfecta, untreated hyperthyroidism, hypogonadism, premature menopause (<45 years), chronic malnutrition, malabsorption and chronic liver disease [31].
Falls were assessed in three different questions
o Current fall resulting in hip fracture (Appendix 4D Q 2) Date and type of fall
o A history of the number/s, date/s and type of fall/s prior to the present fall were recorded in Q3.6
o Elderly Fall Screening Test (EFST) was administrated (Q7)
The EFST has been validated in a number of community studies and was used to assess fall risk. The tool has six items with no objective
testing required, and is therefore easy to implement and has a sensitivity of 93% with a specificity of 78%. The six items include previous falls, lifestyle factors, neurological disorders, medication use and recent illness, balance and gait, and environmental hazards. Subjects who score in more than 3 categories are at high risk for falls [476, 477]. The EFST has not been previously used in SA studies.
3.10.4 Functional assessment
The following scales were used to assess basic and instrumental activity of daily living, quality of life and level of pain:
1. Physical self-maintenance scale (PSMS)
2. Lawton Instrumental Activities of Daily Living scale (IADL) 3. Euro Quality of Life scale (QoL)
4. Oswestry Disability Index (ODI) 5. Visual Analogue Scale (VAS) for pain
The PSMS and the Lawton IADL are widely used in combination to assess daily living activities and are easy to administer. They correlate well with other activity scale scores, with good inter-rater reliability at 0.87 and 0.91 in the elderly [478, 479]. Possible limitations of these scales include under or over reporting as both scales are self-reported and are not observed. Both scales however if administered repeatedly can document functional improvement or deterioration [478, 479]. The
ability to administer the test repeatedly and maintain validity was important in assessing functional outcomes during the study period and at one year. The Lawton IADL scale has been reported to be a sensitive marker of function and correlated with the nutritional index in a SA study [480, 481].
The Euro QoL, a generic questionnaire, used and validated in hip fracture cost studies, enables the comparison between different ethnic and study groups [482]. It is short and easy to administer and provides valuable insight into subjects overall functioning. The Euro QoL has been shown to be sensitive in detecting moderate to severe functional impairment in a multi-ethnic SA population post stroke and in amputees in SA [483-485].
The ODI has been studied extensively in subjects with lower back and hip pain. It has been validated in cross cultural settings, [486, 487] is easy to score and is a good predictor of functional outcome post-surgery. It is scored out of 60 and converted into a percentage. If any section is not answered, the score is adjusted accordingly before the percentage is calculated (Appendix 3.D).
Pain is a significant contributing factor to hip fracture subjects’ recovery. The VAS is a simple tool to administer and has good reliability and validity to measure pain at a particular moment in time. The subject indicates her/his subjective feeling of pain on an unmarked 10cm line and the point is measured and scored out of 10. A score above 4cm predicts a poor outcome and correlates with other functional scales in post hip fracture subjects. It is most useful to determine changes in individuals over time rather than comparing individuals at a discrete point [488, 489].
The ODI and VAS have been used primarily in orthopaedic studies in SA involving lower back pain or recovery after spinal surgery, and although the ODI pain score and VAS pain score may not reveal an initial correlation, with time both scores are considered useful tools for interpreting functional recovery [490, 491].
3.10.5 Physical Examination
All patients had a full physical examination undertaken by a specialist physician (the author) on enrolment to screen for secondary causes for osteoporosis and on subsequent follow up visits to detect any co-morbid illnesses.
3.10.6 Haematological and biochemical investigations
All patients recruited into the case control study had 20 millilitres of venous blood drawn from a vein in the antecubital fossa, for haematological and biochemical evaluation. The investigations were chosen to exclude secondary causes of osteoporosis and to determine if there are haematological or biochemical tests that may help predict risk or outcomes.
The investigations were performed immediately after consent was obtained, and this was usually within a week of admission for hip fracture subjects. The medical file was reviewed prior to performing any tests and any results from admission if
available were documented. Blood investigations were performed in control subjects on the day of recruitment.
The following blood tests were performed on the total cohort:
1. Blood was collected in an ethylenediaminetetraacetic acid (EDTA) tube an analysed in a Sysmex machine (Roche) for a full blood count (FBC). The haemoglobin (Hb) (g/dl), red blood cell mean corpuscular volume (fl) and mean corpuscular haemoglobin (pg), white blood cell count ×109/l (WBC) and platelet count×9/l were measured.
2. Blood samples were collected in a gel tube and analysed in a Beckman Coulter DXC 800 for the following:
a. Urea and electrolyte: Sodium, potassium chloride, bicarbonate, urea and creatinine.
b. Liver function test: Total protein, albumin, bilirubin, alkaline phosphatase (ALP), gamma glutamyl transferase (GGT) and alanine transaminase (ALT).
c. Calcium, phosphate and magnesium: The serum calcium was corrected using the following formula
Corrected calcium = serum calcium + ((40 - albumin) × 0.025) d. Random glucose: (collected in a potassium oxalate tube) e. C- reactive protein (CRP)
3. The Beckman Coulter DXI 600 was used for measuring thyroid function tests and the parathyroid hormone level.
4. 25 hydroxy vitamin D was sent to the Central National Health Service laboratory in Johannesburg for analysis using the high performance chromatography (HBLC) based Chromo-systems diagnostic kit.
5. Oestrogen and testosterone levels were measured using chemical luminescence (Roche).
6. Erythrocyte Sedimentation Rate was measured using the Westergren method using non clotted blood.
All blood samples were processed at the admitting hospital by the National Health Laboratory Service (NHLS). The normal reference ranges for NHLS were used as reference range (Appendix 3.E).
3.10.7 Radiological investigation
3.10.7.1 Conventional radiographs
All hip fracture subjects had hip radiographs on admission. Due to logistic difficulties all subjects were unable to have spine radiographs on admission. Subjects who came for their follow up visits at IALCH had plain thoracic and lumbar spine radiographs performed. Separate antero-posterior (AP) and lateral radiographic views of the spine were acquired using a standardized protocol in 122 hip fracture subjects at their three month visit and 197 control subjects on the day of enrolment.
Of the 78 hip fracture subjects who did not have radiographs, 35 died before their 3 month follow up visit, 29 were unable to come to hospital and were followed up telephonically and 14 were lost to follow up.
A further 4 patients had radiology but did not have a DXA scans due to technical reasons (1 had a recent barium meal, 1 was unable to lie down and two refused).
The comparison (gender and ethnic) between subjects in whom radiology was /was not performed is expanded in section 4.
All radiographs were reported by an experienced specialist radiologist blinded to hip fracture or control subjects, Dr J Maharaj (UKZN, Senior Lecturer in Department of Radiology). Thoracic and lumbar vertebrae were considered to be abnormal (morphometric fracture) using the fixed method i.e. if there was a decrease in height of >20% in its anterior, middle or posterior section compared to its own or nearest intact posterior vertebra [492, 493]. The percentage loss was calculated using the difference in height. Fractures were graded as mild (20-25%), moderate (25.1- 39.9%) or severe (>40%) according to the degree of deformity [10]. This method has been validated in several studies including SOF correlates with clinical measures of height loss, age, back pain, baseline BMD. The measurements are easy to perform and requires no reference range and are accurate even for small sample numbers [183].
3.10.7.2 Bone mineral density
BMD and BMC measurements were obtained using DXA. The Hologic Discovery A densitometer was used to measure BMD of the spine and the opposite hip in hip fracture subjects. The BMD of the lumbar spine (L1 to L4) was obtained and a mean value calculated.
The BMD was measured according to a standardized protocol by trained radiographers, who had previously completed the National Osteoporosis Foundation of South Africa (NOFSA) / International Society for Clinical Densitometry (ICSD) training course, at KEH and IALCH. Phantoms were scanned weekly at each site to ensure reproducibility. A precision assessment tool, which is more accurate than a manual assessment, was used to calculate the percentage coefficient variant (% cv) between the sites. The % cv change was 4.84% across all systems or a change of 0.047 g/cm2 (Appendix 3.F), was considered acceptable for comparison.
BMD was measured in (g/cm2) and was calculated at the hip from BMC (g) divided by bone width (BW) and at the lumbar spine by total scan area. BMD is conventionally expressed as either an absolute value (g/cm2) or a deviation from the norm defined as a Z - score or T - score. The Z - score is a comparison of the subjects BMD with an age, gender and ethnic matched control whilst the T - score is comparison to young adult reference range for that population. For this study, the NHANES III Caucasian data as recommended by the NOFSA [88] was used as there are no normative data for different ethnic groups of SA.
Numerous studies have shown that actual bone size affects areal bone mass measurements especially in different ethnic groups. Several techniques have been proposed to adjust bone mass measurements for volumetric differences in body size.
The volumetric assessment; BMAD was calculated at spine and hip, based on the premise that volume of bone can be calculated from DXA areal measurements as proposed by Carter [494] and Katzman [495] for the lumbar vertebrae and the femoral neck respectively using the formulae:
BMAD = Mean at the lumbar area [494] and
BMAD = B at the femoral neck [495]