Strong Bone Asia 2013 Osteoporosis in ASEAN Lateral Vertebral Analysis DXA Body Composition Quality Assurance in DXA Chris Schultz Scientist-in-Charge (Bone Densitometry) Royal Adelaide Hospital, Adelaide, Australia Copyright 2005-2013 This material is Copyright by the Council of the Australian & New Zealand Bone & Mineral Society (ANZBMS) and may not be copied or transmitted by any means without written permission from the Council of the ANZBMS.
Strong Bone Asia 2013 Osteoporosis in ASEAN Lateral Vertebral Assessment Copyright 2005-2013 This material is Copyright by the Council of the Australian & New Zealand Bone & Mineral Society (ANZBMS) and may not be copied or transmitted by any means without written permission from the Council of the ANZBMS.
Prevalence (%) Vertebral Deformities 80 60 40 20 0 Melton, 1993 55 65 75 86 95 Age (yrs) Common osteoporotic fracture Prevalence with age (10-25% depending on method) More common in women May occur without/minimal trauma often undiagnosed ANZBMS 2005-2013 - Slide 3
Why assess Vertebral deformities? Indicate Impaired bone strength Increased risk of further fractures 1 Up to 4x vertebral, 2x other 20% osteoporotic women fracture again within 12months A need for treatment Increased mortality 2 Up to 6 times with fracture Related to health, but still 2.4x if healthy Entitles the patient to a Medicare rebate for DXA in Australia 7.4 25.1 75 4.4 14.9 Low BMD Med BMD Hi BMD 10.2 Ross PD, Osteopor Int, 1993 1 No Fx 1 Fx > 1 Fx 1. Johnell O, 2001 2. T Jalava JBMR 2003 ANZBMS 2005-2013 - Slide 4
Patients (n) Are Vertebral Fractures Recognized? Only 19% of women with a fracture received treatment 934 women age 60 and older, hospitalized for various reasons chest x-rays reviewed for fracture 140 132 120 100 80 52 % 60 40 20 17% 7% 0 Fracture identified by study radiologists Fracture noted in radiology report Fracture noted in medical record Received osteoporosis treatment S.Gehlbach et al. Osteoporos Int 2000;11:577 ANZBMS 2005-2013 - Slide 5
Where do vertebral fractures occur? T4 T5 T6 T7 T8 T9 T10 T11 T12 L1 L2 L3 L4 Reproduced by kind permission of Sue Steel, Hull, UK 1 2 3 4 5 6 7 8 Spine fracture prevalence (%) Nevitt MC et al. Bone. 1999;25:613 619. Chapurlat R et al. Osteoporos Int. 2006;17:1189 ANZBMS 2005-2013 - Slide 6
Symptoms 1,2 Back pain Loss of stature Reduced mobility in spine Impacts on walking, bending, dressing Functional impairment e.g Pulmonary function Increased bed rest 10x Symptoms are worse with more severe fractures There is a need for improving diagnosis 1. Nevitt MC, 1998 2. Silverman SL, 2001 ANZBMS 2005-2013 - Slide 7
Patients with Symptomatic Vertebral Fractures Not Worth Living In Constant Pain Daily Analgesia Hard to Bend Hard to Stand Wake Early 0 20 40 60 80 % Patients Scane et al, Osteoporosis Int 1994; 4: 89-92. ANZBMS 2005-2013 - Slide 8
Who needs Vertebral Fracture Assessment? Where it might change management See ISCD website Osteopenia + CRFs GCIOP risk Routine vs Selective approach Routine VFA detected 20% prevalence of vertebral fracture in females Selected (targeted) VFA (height loss, Dowager hump, suspected fracture, known VF) only detected 9.6% prevalence Most vertebral fractures (56.2%) occurred in women with osteopenia Routine VFA would be expected to alter the management of 1 in 6 osteopenic women Middleton Osteoporosis Int. 2008 19(8): 1167-73 ANZBMS 2005-2013 - Slide 9
What are the characteristics of Vertebral Fracture? Altered shape Loss of parallelism of the endplates Decreased vertical height (anterior, mid- or posterior) due to vertebral ring fracture Disruption in cortex of central endplate distortion/disruption buckling ANZBMS 2005-2013 - Slide 10
Qualitative Assessment Gold standard is qualitative: Visual diagnosis lateral spine radiograph Dependent on good radiographic techniques This gives fracture: Severity (mild to extreme) Type (wedge, endplate, compression) Newer method (Jiang G, 1994): Algorithm-based Examines central endplate depression ANZBMS 2005-2013 - Slide 11
Quantitative Assessment Deformity = vertebral height Various methodologies: % vertebral height change > 3mm Means & SD of vertebral height ratios Vertebral height normalised for body size using T4 May be automated ANZBMS 2005-2013 - Slide 12
Problems with Quantitative Assessment Reductions in vertebral ht don t only occur in fractures True reductions in ht can only be identified longitudinally Consider changes in endplates and lack of symmetry between vertebrae ANZBMS 2005-2013 - Slide 13
Standardized Visual/Semi-quantitative Assessments Classify/grade using basic features of the vertebral deformities or fractures Shape, type & severity Various techniques Genant methodology is most common No measurements, visual inspection of T4-L4 Good interobserver agreement Mid thoracic and thoraco-lumbar junction normal variants may sometimes be misinterpreted as deformity ANZBMS 2005-2013 - Slide 14
Genant: Mild Deformity Genant, JBMR 1996 ANZBMS 2005-2013 - Slide 15
Genant: Moderate and Severe Deformity Genant, JBMR 1996 ANZBMS 2005-2013 - Slide 16
? Osteoporotic Vertebral Fracture 29 26 29 L3 L2 23 20 23 L2/L3 = 0.79 ANZBMS 2005-2013 - Slide 17
Vertebral Wedge Osteoporotic Fracture Vertebral fractures can be gradual and asymptomatic Diagnosis confounded by end plate irregularities, scoliosis etc Wedge fracture with >20% reduction in A/P heights ANZBMS 2005-2013 - Slide 18
DXA LVA Methods Hologic and Lunar have specific vertebral deformity assessment Morphometric X-ray Absorptiometry (MXA) Instant Vertebral Assessment (IVA, Hologic) Lateral Vertebral Assessment (LVA, Lunar) Others: no dedicated software ISCD: Vertebral Fracture Assessment (VFA) ANZBMS 2005-2013 - Slide 19
Scanning Acquisition Perform a lateral spine scan using the appropriate software Don t use AP spine scan mode Positioning Machines without rotating arms Patient lie on their side using positioning props for stability Machines with rotating arms Patients lie on their backs Arms are raised at right angles to the body or above shoulder height ANZBMS 2005-2013 - Slide 20
Analysis Use same methodology as that used in the reference data Look for deviations from a rectangular shape Measure anterior/posterior/mid heights of: the vertebra of interest the vertebrae immediately above and below If either vertebrae above and below the vertebra of interest appear abnormal in shape, use another vertebra ANZBMS 2005-2013 - Slide 21
H H H p1 m1 a1 H p2 H m2 H a2 Analysis Six points per vertebra Points are in the centre of the vertebral endplate Ratios of heights are determined H p3 H m3 H a3 eg: If L2 is of interest L1 H a1 / H p1» H m1 / H p1 L2 H a2 / H p2» H m2 / H 2 L3 H a3 / H p3» H m3 / H p3 ANZBMS 2005-2013 - Slide 22
Interpretation (1) Compare current to previous scans Perform a qualitative inspection for identifying normal variations and deformities Deformity: Reduction in vertebral height Mild ~ 20-25 % Moderate 25-40 % Severe > 40 % Female 72yr, L2 wedge IVA serial study ANZBMS 2005-2013 - Slide 23
Type of deformity: Interpretation (2) Ant Ht / Post Ht < 0.8 Anterior Deformity Ant Ht / Post Ht > 1.2 Posterior Deformity Mid Ht / Post Ht < 0.8 Concave Deformity ANZBMS 2005-2013 - Slide 24
Interpretation (3) When heights are compared to other vertebrae, e.g. if either (or both, as shown) Ant Ht L2 / Ant Ht L3 Post Ht L2 / Post Ht L3 L2 L3 are < 0.8 Crush deformity ANZBMS 2005-2013 - Slide 25
? Crush Fracture 29 26 29 L2 L3 23 20 23 L2/L3 = 0.79 ANZBMS 2005-2013 - Slide 26
Comparison of methods Good agreement between visual LVA and qualitative radiology Not as good with quantitative alone 1 Sensitivity of LVA / MXA as low as 62% (poor, in comparison with radiography), but specificity high (>90%) 1. Duboeuf F, J Clin Dens, 2005 ANZBMS 2005-2013 - Slide 27
Advantages Lower radiation dose Fast analysis Good agreement between visual LVA and qualitative radiology for moderate and severe deformities Useful for screening Most useful for follow-up of patients with prevalent fractures Evolving technology MXA TM, LVA TM, IVA HD TM, idxa TM ANZBMS 2005-2013 - Slide 28
Disadvantages Lower resolution particularly in thoracic region DXA resolution crude by radiological criteria Doesn t discriminate between types of deformity Less sensitive to mild deformities Not good for scoliotic spines ANZBMS 2005-2013 - Slide 29
Strong Bone Asia 2013 Osteoporosis in ASEAN Body Composition Copyright 2005-2013 This material is Copyright by the Council of the Australian & New Zealand Bone & Mineral Society (ANZBMS) and may not be copied or transmitted by any means without written permission from the Council of the ANZBMS.
Total/Whole Body DXA Scanning Measures Bone: BMD, BMC & Area for entire body Body Composition: divides soft-tissue into Fat (F) & Fat Free Mass (FFM) or lean tissue ANZBMS 2005-2013 - Slide 31
Reasons for Body Composition analysis & Total Body DXA Osteoporosis Paediatrics Anorexia & Obesity Renal failure Malabsorptive disorders (eg Inflammatory bowel disease) Liver disorders (Ascites) Congestive Heart Failure Chronic Obstructive Pulmonary Disease Transplants Cancer & HIV associated wasting Burns & Trauma Spinal Cord Injuries ANZBMS 2005-2013 - Slide 32
Models of Body Composition Multi-compartment models to determine body fat 2C model has many assumptions Add more compartments to overcome limitations Accurate 4C model requires Water, Bone (DXA) & Protein (Neutron activation) Toombs RJ et al, Obesity, 2012 ANZBMS 2005-2013 - Slide 33
Body Weight as Compartments Elemental O + C + H + N + Ca + P + K + Cl + Na + Mg +.. Molecular Fat + Water + Protein + Soft tissue Mineral Cellular Body Cells + Extracellular Water + Extracellular solids + Fat Tissue Adipose + Skeletal Muscle + Bone + Other Ellis K, Phys Rev, 2000 ANZBMS 2005-2013 - Slide 34
Body Composition by Anthropometry Skinfold thickness (2C but incomplete) Simple anthropometry Standard sites Only subcutaneous fat Precision is operator dependent Bioelectrical impedance analysis (2C) Resistance to low current (50Hz & < 1mA) between ankle & wrist Intracellular water vs Extracellular water (ie FFM) Assumes body is simple cylinder, body temp constant, electrolytes uniform Many variables to control (hydration, food intake, activity) Underestimates fat & lean at ends of range Chumlea WC, Baumgartner et al, 1988 ANZBMS 2005-2013 - Slide 35
Body Composition by Element Neutron Activation Analysis (2C or 3C) Neutrons excite atoms & release energy Typically Measure N, Ca (& Cl, Na, P, C) Radiation dose > 3mSv for Ca 0.1mSv for C 0.3mSv for N Precision 1-2% for Ca 3-4% for N Sutcliffe JF, 1996 ANZBMS 2005-2013 - Slide 36
Body Composition by tissue Hydrostatic weighing or air displacement plethysmography (2C) 2 compartment only (F & FFM) Problems with lung & body volume Assumes constant density in FFM compartment errors in assumptions accumulate unless measured Error of fat mass ~4% of body mass ANZBMS 2005-2013 - Slide 37
Body Composition Methods by Molecule Dual Energy X-ray Absorptiometry Becoming the gold standard 3 Compartment in principle Limited by interpolation over bone for compartments ANZBMS 2005-2013 - Slide 38
DXA Body composition Accuracy vs 4C model Accuracy varies with Athlete/Non-Athlete Non athletes GE-Lunar: -1.77% (pencil) & +1.85% (fan) Hologic: -1.6% (pencil) & -3.05% (fan) Athletes GE-Lunar: -1.3% (pencil) & +2.9% (fan) Hologic: -2.1% (pencil) & +2.9% (fan) Better performance in groups than individually Toombs RJ et al, Obesity, 2012 ANZBMS 2005-2013 - Slide 39
DXA Body composition Reproducibility Lean mass (in vivo) GE-Lunar: 1.1% (pencil) 0.85% (fan) Hologic: 0.6% (pencil) 1.3% (fan) Fat mass (in vivo) GE-Lunar: 2.3% (pencil) 1.2% (fan) Hologic: 3.2% (pencil) 1.3% (fan) Toombs RJ et al, Obesity, 2012 ANZBMS 2005-2013 - Slide 40
Scan Analysis Advantage of DXA measurements for soft tissue compared with many other in vivo composition techniques is ability to analyse body segments and make whole-body estimates Mazess et al. 1990 Fuller et al. 1992b Chilibeck et al. 1994 Tothill et al. 1994 Toombs et al 2012 Precision of sub regions is not as good as for the total body tending to 5% for both major manufacturers Different manufacturers offer different degrees of flexibility in the regional analysis but always possible to subdivide the body into at least arms, legs and trunk ANZBMS 2005-2013 - Slide 41
Complications: Artefact removal Artefact removal (GE-Lunar only) The artefact brush ANZBMS 2005-2013 - Slide 42
Total Body BMD ANZBMS 2005-2013 - Slide 43
Soft tissue composition results DXA total body composition reports fat mass (and %fat) lean mass (and %lean) total bone mineral content Regional values android/gynoid region (automatic on GE-Lunar) arms legs trunk Other body composition methods cannot provide regional values for body fat distribution ANZBMS 2005-2013 - Slide 44
Total Body Soft Tissue ANZBMS 2005-2013 - Slide 45
Android/Gynoid Android/Gynoid ROIs Android to Gynoid ratio (waist to hip fat ratio) A measure of obesity which may be closely related to metabolic disturbances and cardiovascular diseases http://dxabodycomp.com/files/oates_article Automatic on GE-Lunar system Android ROI Lower boundary at pelvis cut Upper boundary 20% of distance between pelvis & neck cuts Lateral boundaries are arm cuts Gynoid ROI Upper boundary below Pelvis cut line by 1.5x Android ROI height Gynoid ROI height = 2 x Android ROI height. Lateral boundaries are outer leg cuts ANZBMS 2005-2013 - Slide 46
Scoliosis Quadriplegic Scoliosis and movement Femur prosthesis Geophagy Total Body Difficulties Geophagy, Courtesy Dr MA Laskey, MRC Human Nutrition Research Cambridge ANZBMS 2005-2013 - Slide 47
Limitations of DXA Total Body Scan Physical dimensions of scan table can make it hard to scan tall (>195cm) or wide subjects Weight limitation of scan table (130/150/200kg) can limit use in obese patients Space between scan table and detector with large abdomens Patient soft tissue inaccuracies ANZBMS 2005-2013 - Slide 48
Tall (height 202cm) Obesity Total Body Anatomical variants ANZBMS 2005-2013 - Slide 49
Limitations of DXA Total Body Scan Calibration-manufacturer dependent There can be inconsistent results between different machines from different manufacturers, and even different result between machines from the same manufacturer Software upgrades can change the algorithms that the device uses to calculate body composition Different hardware and software configurations can result in different interpolations of soft tissue over bone, and different treatment of pixels of which a small portion is bone ANZBMS 2005-2013 - Slide 50
Limitations of DXA Total Body Scan Patient soft tissue inaccuracies Assumption that the hydration of fat-free mass is uniform abnormal hydration due to elderly or ill subjects The degree to which body thickness also affects measurements of soft tissue mass: thin and obese patients Pixels that include a small amount of bone may fail to be counted as bone, and appear as very lean tissue because their average absorption coefficient for x-rays is closer to that of lean issue than to bone ANZBMS 2005-2013 - Slide 51
Strong Bone Asia 2013 Osteoporosis in ASEAN Quality Assurance and Quality Control Copyright 2005-2013 This material is Copyright by the Council of the Australian & New Zealand Bone & Mineral Society (ANZBMS) and may not be copied or transmitted by any means without written permission from the Council of the ANZBMS.
Importance of DXA QA and QC To ensure confidence in accurate, reproducible patient results: Monitor densitometer performance Detect densitometer problems quickly identify drift! Achieved with: Scanner dependent procedures Regular scanning of QC phantoms to monitor instrument performance Technologist dependent procedures Operator training and attention to detail Standardised protocols for patient positioning, scan acquisition and scan analysis ANZBMS 2005-2013 - Slide 53
Quality Assurance Should check if densitometer is within engineering specifications for: X-ray system Detector system Measurement of bone & tissue references Pixels (are they square?) GE-Lunar & Norland use compulsory phantom scan Hologic / Norland Use non-compulsory phantom scan ANZBMS 2005-2013 - Slide 54
Quality Assurance Strongly recommended QA be performed at least 3 times per week, & every scanning day Ensure the scanner is maintained with Preventative Maintenance by manufacturer representative regularly (at least every 12 months) Limitation: Only checks individual components. Systematic errors may be cumulative (not detected?) ANZBMS 2005-2013 - Slide 55
Quality Control One aspect of QA that is measurable Provides information on longitudinal variation and stability of the densitometer Involves regularly scanning a phantom of known value and evaluating BMD (and BMC & Area) for change and drift patterns Some systems use this as QA as well ANZBMS 2005-2013 - Slide 56
Strong Bone Asia 2013 Osteoporosis in ASEAN System Quality Assurance Copyright 2005-2013 This material is Copyright by the Council of the Australian & New Zealand Bone & Mineral Society (ANZBMS) and may not be copied or transmitted by any means without written permission from the Council of the ANZBMS.
Instrument Quality Assurance Lunar GE-Lunar QA Components: Scatter gram available no running mean or Shewart charting Calibration is verified using a calibration block phantom containing: Tissue equivalent Shutter material 3 Hydroxy-apatite chambers of varying density (& width on older systems) ANZBMS 2005-2013 - Slide 58
Instrument Quality Assurance Hologic QC/QA software provides: running mean Scatter gram assesses accuracy using Shewart charting Calibration is obtained dynamically using a spinning wheel containing bone soft tissue and air components ANZBMS 2005-2013 - Slide 59
Strong Bone Asia 2013 Osteoporosis in ASEAN Tools for performing Quality Control Copyright 2005-2013 This material is Copyright by the Council of the Australian & New Zealand Bone & Mineral Society (ANZBMS) and may not be copied or transmitted by any means without written permission from the Council of the ANZBMS.
Quality Control Establishing the baseline Provides long term monitoring of densitometer and helps identify drifts or out of control events Monitor QC performance with phantom Acquire baseline scans - either 25 scans (separate days) 10 scans (same day) determine average & standard deviation After this scan at least 3x per week every day a patient is scanned in research studies ANZBMS 2005-2013 - Slide 61
Quality Control Phantom Performing measurements Ensure scanning mode is the same If using GE-Lunar aluminium bar and water bath ensure the water depth (15cm) does not change this is critical Phantom is straight on bed Analyse L1-L4 on all phantoms Use a compare function (ROI copy) when analysing the phantom scans to ensure the regions are the same size as the initial phantom measurement ANZBMS 2005-2013 - Slide 62
Assessing Quality Control Analysis assesses if the scanner is out of control using procedures adapted from manufacturing/engineering Analyse results with: Tabular limits, CV% and scattergrams Shewhart rules applied to scattergram CUSUM charts Running means (moving average) ANZBMS 2005-2013 - Slide 63
Assessing Quality Control Tabular analysis Use baseline phantom scans Determine Average BMD ±1.5% BMD around this control limits Tabulate results can plot with limits = Shewhart chart BMD outside limits? scanner out of control? Trends not identified quickly or clearly ANZBMS 2005-2013 - Slide 64
Assessing Quality Control Scattergrams Graph of phantom BMD vs time Mean ± 1,2 and 3 standard deviations (SD) (calculated from the initial 25 scans) Apply 21 point running mean to data (current average) Identifies drift with scanner - short and long term Statistical variation in plot is likely ANZBMS 2005-2013 - Slide 65
% CV Assessing Quality Control CV% monitoring.100 X Graphical monitoring of the precision over time Recalculate and plot CV of last 10 phantom scans Typically CV will be -1.5% for a phantom ANZBMS 2005-2013 - Slide 66
Assessing Quality Control Shewhart rules Analyses variation Determine Z score of measurement Plot Z score vs time If > 2SD from mean check other rules If multiple rules broken, system out of control ANZBMS 2005-2013 - Slide 67
Assessing Quality Control Shewhart rules Shewhart rules Violation of rules occurs after a 2 SD change in BMD phantom seen and if next BMD phantom exceeds ± 3SD from mean Has 2 consecutive 2SD changes (same side of mean) Has 2 consecutive values differing by more than 4 SD Has 4 consecutive values on same side of mean and > ±1SD Has 10 consecutive values on same side of mean This can be too sensitive often change thresholds and average the next 10 scans (use the running mean) ANZBMS 2005-2013 - Slide 68
Assessing Quality Control CUSUM chart Cumulative sum of the difference between an established mean and the current measurement (g/cm 2 ) a CUSUM BMD b Target Plot CUSUM/δ vs time Need to calculate high and low control limits Deviations outside these are control failures Technically more difficult Example : BMD (Mean 1.125 g/cm 2 ) BMD difference CUSUM scan 1 1.128 0.003 0.003 scan 2 1.121-0.004-0.001 scan 3 1.125 0-0.001 scan 4 1.126 0.001 0 scan 5 1.120-0.005-0.005 b 1 ANZBMS 2005-2013 - Slide 69
Assessing Quality Control CUSUM Ideal results Deviation around zero Identifies time point of change May be too sensitive for DXA ANZBMS 2005-2013 - Slide 70
Assessing Quality Control Moving Averages CUSUM may be overly sensitive Shewhart may not detect changes rapidly Require test sensitive to small change easy to use Moving average smooths out small daily change more data points increase sensitivity, but slow response ANZBMS 2005-2013 - Slide 71
Assessing Quality Control Simple Moving average Smooths out variation in tabular value or ( t k ) ( t 1) BMD R R 2 BMD t BMD k 1 t k 1 Normally, k = 20 (or 10 in some cases) The running mean value should vary randomly around the mean ( 1 SD) Check BMD within limits as before May miss significant trends ( t k ) ( t k ) 2 BMD ANZBMS 2005-2013 - Slide 72
Assessing Quality Control Simple moving average applied to Scattergrams ANZBMS 2005-2013 - Slide 73
Assessing Quality Control Simple moving average applied to Scattergrams ANZBMS 2005-2013 - Slide 74
Assessing Quality Control Weighted Moving Average Variation on running mean Weights for each term Applies more weight to most recent data MA BMD t ( 1. BMD t 1 (2) BMD t 2 (3) BMD t 3 k( k 1) / 2... ( k) BMD t k ) A simple moving average if all weights = 1 ANZBMS 2005-2013 - Slide 75
Weighted Moving Average ANZBMS 2005-2013 - Slide 76
Assessing Quality Control Exponentially Weighted Moving Average Weights are exponentially changing More rapid detection of change Require smoothing factor from 0 1 2 k 1 Estimate where k = number points As k increases, EWMA tends to 0 1 is a Shewhart plot Normally 0.15 For 20 points 0.09 ANZBMS 2005-2013 - Slide 77
Exponentially Weighted Moving Average ANZBMS 2005-2013 - Slide 79
Strong Bone Asia 2013 Osteoporosis in ASEAN What if the QC fails? Copyright 2005-2013 This material is Copyright by the Council of the Australian & New Zealand Bone & Mineral Society (ANZBMS) and may not be copied or transmitted by any means without written permission from the Council of the ANZBMS.
Resolving QC failure Single measurement failure : repeat QC and QA if passes most likely statistical variance if fails call service representative Running mean exceeds 1SD most likely scanner shift call service representative ANZBMS 2005-2013 - Slide 81
Resolving QC failure May be due to: Relocation of densitometer service events variations in temperature, humidity, power, ageing part replacement X-ray tube, detector or electronics failure ANZBMS 2005-2013 - Slide 82
Scanner Maintenance and Repair DXA scanners require regular preventive maintenance Record all maintenance and repairs Recalibration may be necessary after repairing or relocating equipment ANZBMS 2005-2013 - Slide 83
My machine is out of control! Re-establishing QC baseline After major changes such as X-ray tube or detector replacement New baseline for phantom must be established Scans prior to failure may need to be verified for accuracy by re-analysis or re-acquisition ANZBMS 2005-2013 - Slide 84
Densitometer Breakdowns Assessing function after breakdown Minimum of 10 phantom scans after repair Mean & Standard Deviation (SD) calculated Stable period (from running mean) prior to breakdown New scans Percent difference determined Change should be less than 2SD (~1%) ANZBMS 2005-2013 - Slide 85
What does all this mean? Performance is monitored and understood Action taken immediately to rectify problems and ensure reliable results Patient results are not compromised by incomplete knowledge of DXA machine performance Best clinical practice! ANZBMS 2005-2013 - Slide 86