Dr T. R. Coe Department of Anaesthesia, Queen Elizabeth Hospital, Egdbaston, Birmingham B15 2TH, UK
Four observers estimated the weight and height of 38 patients who were lying covered on operating theatre trolleys. Regression analysis of the results was performed for each observer. The results showed marked variation in ability to assess these characteristics accurately.
It is important for an anaesthetist to know a patient's weight in order to give an accurate drug dosage. This is particularly important in a sick patient, when the administration of an excessive dose of induction agent might be unsafe or potentially life threatening. Similarly, an inconveniently long neuromuscular block may result if a patient is given a dose of neuromuscular blocker that is too large. Height measurement is required less frequently but will be needed for calculations involving body surface area, e.g. cardiac index and body mass index determination.
The majority of patients can be weighed and measured before anaesthesia. Alternatively, patients can be questioned, in which circumstances they tend to give reasonably accurate estimates of their own height and weight . However, some patients, such as elderly trauma patients or intensive care patients, are often unable to be measured or to give estimates because of decreasing mental faculty, immobility or unconsciousness. The failure to obtain and record weight and height has been highlighted in the National Confidential Enquiry into Peri-operative Deaths (NCEPOD) reports. One report concluded that this might reflect poor quality of care . It goes on to suggest that the anaesthetist's assessment of weight is reasonable. This assumption has not been questioned or quantified. This study aimed to evaluate the ability of anaesthetic staff to assess patients' weight and height.
Thirty-eight patients undergoing day case orthopaedic or urological surgery were included in this study. The Chairman of the Hospital Ethics Committee waived the need for patient consent. The patients were assessed in the anaesthetic room on identical trolleys, wearing a standard hospital gown and covered by a single blanket. The weight and height of each patient was visually assessed by the four investigators: a consultant with 18 years' anaesthetic experience, a staff grade with 9 years' anaesthetic experience, a senior house officer with 5 months' anaesthetic experience and an operating department assistant with 11 years' experience. None of the patients was seen before lying on the trolley and the actual weight and height of each patient was not made known to any estimator until the end of the study. Anaesthesia was given by target controlled infusion of propofol, the weight of the patient being entered into the pump by an independent anaesthetic assistant.
Regression analyses were performed for height and weight estimation for each observer, using the 38 observations in each case. The estimate from the member of staff is used as the explanatory variable and from this it is hoped to estimate the true weight or height. A predictive equation is therefore specific to a particular member of staff. The results are shown graphically with a plot of the regression line together with the predictive equation and r2 value. In the predictive equation, x is the estimated variable and y is the derived ‘true’ weight or height. The r2 value shows how well the line fits the data in each case, a value of 1.0 implying a perfect fit and 0 implying no fit at all.
Thirty-eight patients were assessed by all four investigators. The characteristics of the patients are summarised in Table 1.
Table 1. Patient characteristics. Values are given as mean (range) where appropriate.
The scatter diagrams and regression lines for each grade of observer are shown in Fig. 1 for weight and Fig. 2 for height. The lines of best fit for the consultant's and operating department assistant's estimations of weight and all but the senior house officer's height estimation pass through the origin. The best fit of the other regression lines produced lines that cross the y-axis at a negative y-value.
Testing the error in weight estimation between observers using analysis of variance showed that the senior house officer consistently overestimated, making his estimations significantly worse (p < 0.001) when compared with the other three estimators. Similar testing for height data showed that the errors between observers were not significantly different (p 0.05) and that there was a tendency for all observers to overestimate. The results expressed as percentage error for each actual weight or height are shown as scatter diagrams in Figs 3 and 4.
The results of our study show that each observer has a different ability, with specific regression lines for estimation of height and weight. The most consistently accurate estimators of weight were the consultant and operating department assistant, with regression lines reasonably close to identity. The regression lines for both the senior house officer and the staff grade have steeper gradients. The staff grade equation, nevertheless, does give useful information, showing overestimation at low weight and underestimation at high weight, with the crossover point at the approximate weight of that investigator. It is possible that one method of estimating weight is to compare the patient's physique with one's own, thereby being more accurate at similar weights but not at weights that are greatly different. However, the equation for the senior house officer shows overestimation at weights less than 100 kg, his own weight being considerably less than this.
With height estimation there was considerably more scatter in the results, particularly the estimations by the operating department assistant and senior house officer. The latter tended to overestimate smaller heights and underestimate taller heights, the crossover being about 1.79 m, somewhat taller than that author. The other three investigators all consistently tended to overestimate height.
One important factor when considering the accuracy of estimation is the amount of error that is acceptable in clinical practice. Weight is important in anaesthesia, as drugs are given on a dose per unit weight basis. However, even a 20% error in adults would probably not produce significant clinical problems, even with neuromuscular blockade. From the percentage error scatter diagrams, it can be seen that most of the estimations are within ± 20% error. Of course, actual body weight is less important than lean body mass when calculating drug doses and underestimation of weight in overweight patients may not be such a bad thing. However, it is important to know if one consistently overestimates at lower than normal body weight, as drug dosage in this situation may represent a significant overdose. This highlights the need for titrating dose and response, particularly at extremes of body weight. Consistent underestimation in overweight patients might suggest subconscious estimation of lean body weight rather than actual weight. However, we were consciously trying to estimate actual, rather than lean, weight. Indeed, the consistent overestimation in lean lower weight patients would bear out this fact.
Although measurement of height is widely used, it is rarely needed in the context of day care surgery. However, estimation of a patient's height is occasionally needed for cardiac output studies when the anaesthetist does not have easy access to the whole of the patient's body or does not have a tape measure. In our study, the range of heights was from 152 to 185 cm. A 20% error in estimation would give rise to clinically significant errors in the results. From the percentage error scatter diagrams, it can be seen that the majority of estimations fit within ± 5% error and this error margin would probably be acceptable. The major discrepancy appears to be overestimation of height in smaller patients.
Height estimation is more difficult in the supine than the upright position. We deliberately did not see the patients until they were lying down, as this would have biased the estimations. Supine height is probably estimated in relation to the length of the trolley and this would explain the smaller error margin at heights closer to the trolley length. If a tape measure is available and access to the whole patient is denied, as may be the case during an operation, height can be estimated from femur length  or lower leg length , assuming you have the appropriate regression equations and access to the relevant body part.
In our study, better overall accuracy of estimation of weight was achieved by the more experienced authors. This may result from more time spent assessing weights and unconsciously correlating them with the visual images of patients. Anaesthesia is probably the only speciality in which a patient's weight is noted and used frequently. The senior house officer author was significantly worse at weight estimation than the other three authors. This may be accounted for by the short exposure time to using drug doses per unit weight. However, the staff grade with 9 years' experience was also inaccurate. Size of estimator appeared to have no influence on accuracy or pattern of estimation as all the authors are of average build.
In conclusion, it would appear that the accuracy of estimation of patient weight and height varies from observer to observer, with everyone having a unique ability. It is possible that improvement in ability could be achieved after elucidating personal trends in estimation with regression analysis. The study shows that our estimations may not be as reasonable as NCEPOD suggest and caution should be exercised when using the guesses particularly at the extremes of physical size.
We thank Hilary Sanders of the University of Plymouth for help with the statistical analysis.