BMJ 1996;313:106 (13 July)

Education and debate

Statistics Notes: Measurement error proportional to the mean

J Martin Bland, professor of medical statistics,^a Douglas G Altman, head ^b

^a Department of Public Health Sciences, St George's Hospital Medical School, London SW17 0RE, ^b ICRF Medical Statistics Group, Centre for Statistics in Medicine, Institute of Health Sciences, PO Box 777, Oxford OX3 7LF

Correspondence to: Professor Bland.

We often need to know the error with which measurements are made--for example, so that we can decide whether the change in a clinical observation represents a real change in a patient's condition. We have discussed previously the within-subject standard deviation as a practical index of measurement error.¹ We said that this approach should be used when the measurement error was not related to the magnitude of the measurement and recommended that we plot the subject standard deviation against the subject mean to check this. Table 1 shows some duplicate salivary cotinine measurements taken from a larger study. Figure 1 shows absolute subject difference against subject mean, which is equivalent to a standard deviation versus mean plot when we have only two measurements per subject.¹ If we are to use the within-subject standard deviation as an index of measurement error we need the subject standard deviation to be independent of the subject mean. Here, there is a clear relationship, the variability increasing with the magnitude. We can test this using a rank correlation coefficient if we wish; here Kendall's (tau) = 0.62, P = 0.0001. Under these circumstances a logarithmic transformation of the data almost always solves the problem, but we can check by plotting log standard deviation against log mean. For these data the slope is 0.9; as this is very close to 1 the subject standard deviation is roughly proportional to the subject mean and a log transformation is indicated.² Figure 2 shows the plot of absolute difference versus subject mean for the log transformed data. There is now no evidence of a relationship (Kendall's (tau) = 0.07, P = 0.7).

Table 1--Duplicate salivary cotinine measurements for a group of Scottish schoolchildren (ng/ml) (D Strachan, personal communication) ---------------------------------------------------------- Subject Measurement Subject Measurement No lst 2nd No 1st 2nd ---------------------------------------------------------- 1 0.1 0.1 11 1.2 0.9 2 0.2 0.1 12 1.9 2.8 3 0.2 0.3 13 2.0 1.4 4 0.3 0.4 14 2.7 1.4 5 0.3 0.4 15 2.8 6.8 6 0.4 0.3 16 3.2 2.9 7 0.4 1.4 17 4.7 4.5 8 0.8 0.5 18 4.9 1.4 9 1.0 1.6 19 4.9 3.9 10 1.1 0.9 20 7.0 4.0

View larger version (21K): [in this window] [in a new window]   Fig 1--Absolute difference against mean for data in table 1

View larger version (19K): [in this window] [in a new window]   Fig 2--Absolute difference against mean after log10 transformation

As the variability is now independent of the magnitude of the measurement, we can calculate the within-subject standard deviation¹ as (sigma)w = 0.175. This is a standard deviation on the logarithmic scale, so we need to antilog it before we can interpret it easily. We will denote the antilog of (sigma)w by (alpha)(sigmaw). For the cotinine data, (alpha)(sigmaw) = 1.496.

When we antilog (sigma)w we have a ratio, not a quantity measured in the units of the original data. This is because to calculate a standard deviation we subtract the mean from each observation. Subtracting on the logarithmic scale is equivalent to dividing on the natural scale.³ Dividing the observation by the mean in this way produces a dimensionless ratio. Hence (alpha)(sigmaw) is not a standard deviation in the original units, but a related quantity sometimes referred to as the geometric standard deviation.

To estimate one standard deviation on either side of the observed value, we should multiply and divide by (alpha)(sigmaw). The difference between a subject's measurement and the true value would be expected to be less than 1.96(sigma)w for 95% of observations.¹ To get the equivalent of 1.96 log scale standard deviations on either side of an observed value we would multiply and divide on the natural scale by (alpha)^1.96(sigmaw) or approximately (alpha)²(sigmaw). This procedure gives limits which should include the subject's mean for 95% of observations. Thus for the cotinine data we would divide and multiply by 1.496² = 2.238. A measurement of 2 ng/ml would tell us that the person's true value probably lies somewhere between 2/2.238 = 0.9 and 2 x 2.238 = 4.5 ng/ml.

Multiplying on the natural scale is equivalent to adding on the log scale. Multiplying a subject's actual measurement by (alpha)(sigmaw) is equivalent to adding one standard deviation on the log scale. Provided the standard deviation is not large compared with the level of the measurement, (alpha)(sigmaw)-1 is approximately equal to the standard deviation expressed as a proportion of the measurement. The ratio of standard deviation to mean is called a coefficient of variation, and here (alpha)(sigmaw)-1 is the within-subject coefficient of variation.

For the cotinine data the estimated coefficient of variation is 1.496-1 = 0.496 or 49.6%. This is rather too large for the approximation to be reliable.

The within-subject variability for salivary cotinine seems very large, but the possible range of values, from these very lightly exposed children to heavy smokers, is very wide and salivary cotinine is sufficiently precise to distinguish between many different levels of exposure. The precision of a method of measurement must be interpreted in the light of its intended use.

1. Bland JM, Altman DG. Measurement error and correlation coefficients. BMJ 1996;313:41-2. [Free Full Text]
2. Bland JM, Altman DG. Transforming data. BMJ 1996;312:770. [Free Full Text]
3. Bland JM, Altman DG. Logarithms. BMJ 1996;312:700. [Free Full Text]

This article has been cited by other articles:

• Verburg, B. O., Jaddoe, V. W.V., Wladimiroff, J. W., Hofman, A., Witteman, J. C.M., Steegers, E. A.P. (2008). Fetal Hemodynamic Adaptive Changes Related to Intrauterine Growth: The Generation R Study. Circulation 117: 649-659 [Abstract] [Full text]  
• Riordan, M. M., Kovacs, S. J. (2007). Absence of diastolic mitral annular oscillations is a marker for relaxation-related diastolic dysfunction. Am. J. Physiol. Heart Circ. Physiol. 292: H2952-H2958 [Abstract] [Full text]  
• DeLeon Ortega, J. E., Sakata, L. M., Kakati, B., McGwin, G. Jr, Monheit, B. E., Arthur, S. N., Girkin, C. A. (2007). Effect of Glaucomatous Damage on Repeatability of Confocal Scanning Laser Ophthalmoscope, Scanning Laser Polarimetry, and Optical Coherence Tomography. IOVS 48: 1156-1163 [Abstract] [Full text]  
• Khambay, B. S., McHugh, S., Millett, D. T. (2006). Magnitude and reproducibility of forces generated by clinicians during laceback placement. J. Orthod. 33: 270-275 [Abstract] [Full text]  
• Ng, Q. S., Goh, V., Klotz, E., Fichte, H., Saunders, M. I., Hoskin, P. J., Padhani, A. R. (2006). Quantitative Assessment of Lung Cancer Perfusion Using MDCT: Does Measurement Reproducibility Improve with Greater Tumor Volume Coverage?. Am. J. Roentgenol. 187: 1079-1084 [Abstract] [Full text]  
• Riordan, M. M., Kovacs, S. J. (2006). Relationship of pulmonary vein flow to left ventricular short-axis epicardial displacement in diastole: model-based prediction with in vivo validation. Am. J. Physiol. Heart Circ. Physiol. 291: H1210-H1215 [Abstract] [Full text]  
• Nordin, E., Rosendahl, E., Lundin-Olsson, L. (2006). Timed "Up & Go" Test: Reliability in Older People Dependent in Activities of Daily Living-- Focus on Cognitive State. ptjournal 86: 646-655 [Abstract] [Full text]  
• Muller, M.J., Mazanek, M., Weibrich, C., Dellani, P.R., Stoeter, P., Fellgiebel, A. (2006). Distribution characteristics, reproducibility, and precision of region of interest-based hippocampal diffusion tensor imaging measures.. Am. J. Neuroradiol. 27: 440-446 [Abstract] [Full text]  
• Lodge, M. A., Badawi, R. D., Gilbert, R., Dibos, P. E., Line, B. R. (2006). Comparison of 2-Dimensional and 3-Dimensional Acquisition for 18F-FDG PET Oncology Studies Performed on an LSO-Based Scanner. JNM 47: 23-31 [Abstract] [Full text]  
• Riordan, M. M., Kovacs, S. J. (2006). Quantitation of mitral annular oscillations and longitudinal "ringing" of the left ventricle: a new window into longitudinal diastolic function. J. Appl. Physiol. 100: 112-119 [Abstract] [Full text]  
• Annet, L., Peeters, F., Horsmans, Y., Hermoye, L., Starkel, P., Van Beers, B. E. (2005). Esophageal Varices: Evaluation with Transesophageal MR Imaging--Initial Experience. Radiology 238: 167-175 [Abstract] [Full text]  
• Hickman, S. J., Wheeler-Kingshott, C. A. M., Jones, S. J., Miszkiel, K. A., Barker, G. J., Plant, G. T., Miller, D. H. (2005). Optic Nerve Diffusion Measurement from Diffusion-Weighted Imaging in Optic Neuritis. Am. J. Neuroradiol. 26: 951-956 [Abstract] [Full text]  
• Hermoye, L., Laamari-Azjal, I., Cao, Z., Annet, L., Lerut, J., Dawant, B. M., Van Beers, B. E. (2005). Liver Segmentation in Living Liver Transplant Donors: Comparison of Semiautomatic and Manual Methods. Radiology 234: 171-178 [Abstract] [Full text]  
• KELLY, F. J., LEE, R., MUDWAY, I. S. (2004). Inter- and Intra-Individual Vitamin E Uptake in Healthy Subjects Is Highly Repeatable across a Wide Supplementation Dose Range. Ann. N. Y. Acad. Sci. 1031: 22-39 [Abstract] [Full text]  
• Wei, X., Guttmann, C. R., Warfield, S. K, Eliasziw, M., Mitchell, J R. (2004). Has your patient's multiple sclerosis lesion burden or brain atrophy actually changed?. Mult Scler 10: 402-406 [Abstract]  
• Revel, M.-P., Lefort, C., Bissery, A., Bienvenu, M., Aycard, L., Chatellier, G., Frija, G. (2004). Pulmonary Nodules: Preliminary Experience with Three-dimensional Evaluation. Radiology 231: 459-466 [Abstract] [Full text]  
• Fiorella, D., Heiserman, J., Prenger, E., Partovi, S. (2004). Assessment of the Reproducibility of Postprocessing Dynamic CT Perfusion Data. Am. J. Neuroradiol. 25: 97-107 [Abstract] [Full text]  
• Hickman, S J, Kapoor, R, Jones, S J, Altmann, D R, Plant, G T, Miller, D H (2003). Corticosteroids do not prevent optic nerve atrophy following optic neuritis. J. Neurol. Neurosurg. Psychiatry 74: 1139-1141 [Full text]  
• Galbraith, S. M., Maxwell, R. J., Lodge, M. A., Tozer, G. M., Wilson, J., Taylor, N. J., Stirling, J. J., Sena, L., Padhani, A. R., Rustin, G. J.S. (2003). Combretastatin A4 Phosphate Has Tumor Antivascular Activity in Rat and Man as Demonstrated by Dynamic Magnetic Resonance Imaging. JCO 21: 2831-2842 [Abstract] [Full text]  
• Stumpf, K., Adlercreutz, H. (2003). Short-Term Variations in Enterolactone in Serum, 24-Hour Urine, and Spot Urine and Relationship with Enterolactone Concentrations. Clin. Chem. 49: 178-181 [Full text]  
• Valentine, J., Simpson, J. M (2002). Deciding whether a change in a patient's performance is a genuine change. Clin Rehabil 16: 837-842 [Abstract]  
• Green, J., Forster, A., Young, J. (2002). Reliability of gait speed measured by a timed walking test in patients one year after stroke. Clin Rehabil 16: 306-314 [Abstract]  
• Reinmann, B., Stocks, J., Frey, U. (2001). Assessment of an infant whole-body plethysmograph using an infant lung function model. Eur Respir J 17: 765-772 [Abstract] [Full text]  
• Gøtzsche, P. C, Hammarquist, C., Burr, M. (1998). House dust mite control measures in the management of asthma: meta-analysis. BMJ 317: 1105-1110 [Abstract] [Full text]  
• Gawne-Cain, M L, O'Riordan, J I, Coles, A, Newell, B, Thompson, A J, Miller, D H (1998). MRI lesion volume measurement in multiple sclerosis and its correlation with disability: a comparison of fast fluid attenuated inversion recovery (fFLAIR) and spin echo sequences. J. Neurol. Neurosurg. Psychiatry 64: 197-203 [Abstract] [Full text]  
• Salinas, M., Rosas, J., Iborra, J., Manero, H., Pascual, E. (1997). Comparison of manual and automated cell counts in EDTA preserved synovial fluids. Storage has little influence on the results. Ann Rheum Dis 56: 622-626 [Abstract] [Full text]  
• Masse, J., Bland, J M, Doyle, J R, Doyle, J M (1997). Measurement error. BMJ 314: 147-147 [Full text]