Published ahead of print on July 17, 2003, doi:10.1164/rccm.200303-451OC
© 2003 American Thoracic Society
Prediction of Childhood Pulmonary Function Using Ulna LengthDepartment of Respiratory Medicine, Royal Children's Hospital, Parkville, Victoria; and Department of Paediatrics, University of Melbourne, Melbourne, Australia Correspondence and requests for reprints should be addressed to Leanne M. Gauld, M.B.B.S., F.R.A.C.P., Respiratory Laboratory, Sydney Children's Hospital, High Street, Randwick, New South Wales 2031, Australia. E-mail: lmgabk{at}hotmail.com
Pulmonary function is important in neuromuscular weakness. In children, height determines normal values. Height measurement is unreliable when neuromuscular weakness or spinal deformity is present. The aim of this study was to accurately predict pulmonary function from a limb segment measurement that is precise and reproducible. Normal males (n = 1,144) and females (n = 1,199), 5.3 to 19.6 years old, were recruited from Melbourne schools. Height, weight, ulna, forearm, tibia, and lower leg lengths were measured using a Harpenden stadiometer and calipers, and electronic scales. Three maximal expiratory maneuvers were performed. Limb measurements were highly reproducible. Linear regression on log-transformed FEV1 and FVC was used to develop prediction equations from limb measurements and age. In males FEV1 = exp (0.071 x U + 0.046 x A - 1.269), r2 = 0.86; FVC = exp (0.77 x U + 0.041 x A - 1.285), r2 = 0.86 and in females FEV1 = exp (0.072 x U + 0.041 x A - 1.272), r2 = 0.84; FVC = exp (0.078 x U + 0.037 x A - 1.315), r2 = 0.83 (U refers to ulna length and A refers to age). Precision is similar to equations using height. Ulna measurement is accessible in wheelchair-bound children. Using ulna length to predict pulmonary function should facilitate respiratory assessment in children whose height is difficult to measure.
Key Words: pediatrics • anthropometry • respiratory function tests • neuromuscular diseases • scoliosis The pulmonary consequences of neuromuscular weakness result from impaired function of respiratory muscles and spinal deformity (1, 2). The impact on pulmonary function depends on the pattern of involvement and the rate of progression of the underlying disease (3–6). Outcome is related to the rate of decline in pulmonary function (6). Monitoring pulmonary function is important in neuromuscular weakness and spinal deformity, where restrictive defects occur and major surgery, such as corrective spinal surgery, is required (7–10). Pulmonary function changes throughout childhood and is related to height and age. Height has traditionally been used to predict normal values (11–14). When spinal deformity, weakness, or immobility is present, height measurement is difficult and inaccurate. Arm span has been used to estimate height, and prediction equations have been developed for healthy children in various populations (15–19). Precision of arm span measurements is limited by weakness and joint deformity that restrict the ability to actively extend the arms fully. Spinal deformity alters the position adopted during the measurement and reduces accuracy. It is common practice to predict height from a measurement of arm span obtained with a flexible tape measure that is run over the skin and around corners. This practice leads to significant error and has poor reproducibility. This study aimed to identify a long bone or distal body segment measurement that could be precisely and reproducibly measured and could be used to accurately predict measurements of pulmonary function.
Subjects A total of 27 of 40 approached metropolitan Melbourne schools were recruited using a presentation at a routine school assembly, and students aged 7 to 18 years were invited to participate. Interested students received an information sheet, a questionnaire, and a consent form for their parent or guardian to read and complete. The questionnaire asked for details on gender, date of birth, medical information, gestation at birth, medication use, and racial background. Students were excluded if the questionnaire revealed significant respiratory or systemic illness, spinal deformity, disease known to cause growth disturbance, prematurity (< 35 weeks gestation), muscle weakness or abnormal tone, use of medications thought to alter growth, or use of asthma preventer or reliever medication in the preceding 4 weeks.
Measurements Distal limb measurements were chosen because they are the most accessible in immobilized children. Although the upper arm is also accessible, landmarks are more difficult to define, which impedes reproducibility (18). Arm span was included so that comparison could be made with the best currently available method of height estimation in this group. The Harpenden stadiometer and calipers and Wedderburn electronic scales were used because of their high degree of accuracy. Each student underwent a brief medical examination of the spine for scoliosis. Those in whom scoliosis was thought likely were referred to their local doctor and were excluded from further participation. Standing height was obtained using a portable Harpenden Stadiometer (Holtain Ltd, Crosswell, UK). Each student placed his/her feet on the base plate, together throughout their length. The back was placed firmly against the vertical plate, with the head, shoulders, buttocks, and heels touching this plate. The head was placed in the Frankfurt horizontal plane. The head plate was lowered to sit on the vertex. Gentle vertical traction was applied to the mastoid processes as a deep inspiration was taken to obtain the measurement. Weight was obtained by standing on zeroed Wedderburn electronic scales (Wedderburn, Southampton, UK). Weight measurement was omitted in the presence of plaster casts and in the case of one student who weighed in excess of the scales ability to measure (150 kg). Arm span was measured on a wooden arm span stadiometer (17). The frame consisted of two vertical poles and a horizontal beam with measurements marked in millimeters. The frame sits neatly against the wall, and each pole is held stable by three wooden feet. Attached to the horizontal beam is a mobile vertical wooden plate that is positioned 80 to 200 cm above the floor. Students were measured while standing with their back against the wall, with their head, shoulders, buttocks, and heels touching the wall. The feet were vertically below the head, which was in the Frankfurt horizontal plane. The feet were together throughout their length. The arms were extended laterally so that the hands were at the same level from the floor as the shoulders. The palms faced forward. The middle finger of the right hand just touched the protruding right hand pole. The vertical plate was moved medially until it rested against the middle finger of the left hand. If the middle finger had been traumatically amputated or injured, this measurement was omitted. Ulna length was obtained in the sitting position with the left forearm resting comfortably on a table. The palm faced downward, and the fingers were extended but together. The elbow was bent at 90 to 110°. The proximal end of the ulna was found by palpating along its length. The tip of the styloid process was felt at the wrist by palpating down the length of the bone distally, until its end was felt. The tips of Harpenden calipers were placed adjacent to both end points (see Figure 1) .
Forearm measurements were made in the same position as the ulna measurements. Harpenden calipers were used to measure from the tip of the left middle finger to the most lateral aspect of the elbow. If the middle finger of the left hand had been traumatically amputated or injured, the right arm was substituted. Tibia length was measured in the seated position, using Harpenden calipers. The right tibial plate was found by palpating along the medial border of the tibia until its proximal aspect was identified. The distal end was found by palpating the most distal point of the medial malleolus. Lower leg length was measured by the Harpenden portable stadiometer. The student was seated beside the stadiometer, facing it. The level of the seat was set such that the knee was at a higher level than the hip. The lateral aspect of the left lower leg was placed firmly against the stadiometer. The foot was placed directly below the knee. The head plate of the stadiometer was brought down to rest on the upper aspect of the knee. To determine reproducibility of the measurements, 14 subjects were measured on two separate occasions on the same day by L.M.G. Both L.M.G. and an independent observer, trained to carry out the measurements, measured another 15 subjects. The first measurement obtained by L.M.G. was included in the analysis for the prediction of pulmonary function on all occasions. Subjects underwent pulmonary function tests (PFT) in the standing position with a nose clip. A Jaeger Masterscope Spirometer with a heated pneumotac and Version 3.43 Jaeger software (Jaeger Toennies, Hoechberg, Germany) was used. Children were asked to take a maximal inspiration, followed immediately by a maximal forced expiratory maneuver without a pause in between. A minimum of three maneuvers was performed. Maneuvers were considered acceptable if there was a rapid rise in peak flow and a full maximally prolonged expiratory curve (20). Computerized visual incentives were often used for encouragement. To ensure reproducibility, FEV1 and FVC of two maneuvers were required to be within 200 ml of each other. The best maneuver was defined as the one with the greatest sum of FEV1 and FVC and was used for analysis (20). FEV1, FVC, and mean midexpiratory flow (MMEF) were recorded. If acceptable and reproducible PFT could not be obtained after several attempts, PFT were not included in the analysis. To evaluate the feasibility of performing the limb measurements in children with neuromuscular weakness, each measurement was performed on 20 children with Duchenne Muscular Dystrophy (DMD). A subjective evaluation of the ease of positioning, identification of landmarks, and performance of the measurements was made.
Statistical Analysis Data were double entered into two identical forms of a Microsoft Access database that had built-in validation checks. Stata statistical software was used for analysis (22). Exploratory analysis indicated that linear regression models performed best on logarithmically transformed pulmonary function values. Linear regression equations were developed including each of the anthropometric measurements and age. Reference ranges were created by back transforming 95% prediction intervals obtained from the regression models; because linear prediction was performed in the log scale, back transformation automatically produces a range that can be expressed in ratios ("% predicted") relative to the central predicted or normal values. The white and Asian subgroups were compared with the overall group by using interaction terms in each linear regression analysis to assess whether the pattern of the relationship between the dependent measure (pulmonary function) and the two independent measures of age and anthropometric measurement was similar between the two groups. A two-degree-of-freedom Wald test was used, and separate prediction equations were obtained whenever this test indicated statistical difference at the 0.05 level. Intra- and interobserver variability was analyzed by calculating the SD of differences between repeated measurements by the same observer (intra) and independent observers (inter) (23). These were used to derive the SD of individual values, which was expressed as a percentage of the mean.
Ethical Approval
A total of 2,810 students (53% of available students) returned a completed questionnaire and a consent form. The response rate was 71% for primary school students and 33% for high school students. A total of 467 students were excluded on the basis of their questionnaire or scoliosis check. The reasons for exclusion are shown in Table 1 . The ages of included subjects were distributed equally from 7 to 18 years, with few lying outside this range.
The measurement of each variable was highly reproducible. The SD of intraobserver differences for repeat ulna measurements was 0.13 cm, corresponding to an SD of 0.093 cm for individual values or 0.41% relative to the mean value. Intra- and interobserver variability was calculated similarly for all repeat measurements, and results are shown in Table 2 .
Prediction equations for FVC, FEV1, and MMEF were developed using each of the anthropometric measurements as predictor variables and are recorded in Tables E1–E6 in the online supplement. Weight did not add to the precision of prediction equations. The accuracy of prediction of FEV1 and FVC from all measurements, including height, arm span, and ulna lengths is comparable. The r2 ranged between 0.85 and 0.89 for males and between 0.81 and 0.88 for females. The prediction equations developed for each of the pulmonary function parameters using ulna length and age are shown in Table 3 together with r2, RMSE, and 95% reference ranges. The relationship between FVC and ulna length is graphically represented in Figure 2 , ignoring age. The prediction equations depicted had r2 of 0.84 and 0.79, respectively, compared with the values of 0.86 and 0.83 when age is included. An example of the use of the prediction equations using the ulna length to FVC prediction is provided in Figure E1 in the online supplement.
The majority of children were white (n = 1,801) or Asian (n = 307). The most anthropometrically distinct subpopulation was the group from Asia. Prediction equations for FEV1 and FVC were significantly different in the female Asian subgroup when using arm span, ulna length, and forearm length. The order of the difference in the prediction from ulna measurement was approximately a 10% overestimation of both FEV1 and FVC. The prediction equations derived for the Asian subgroup, their r2, RMSE, and reference ranges are included in Table 4 . Subgroup analysis for the other measurements and for the prediction of MMEF did not uncover significant differences. All the developed prediction equations were representative of the white population.
Twenty children with DMD underwent the limb measurements. The mean age was 12.1 years (7.2–18.6), and 17 children were wheelchair bound. All children had significant fixed ankle deformity, and 14 had significant wrist and elbow contractures. The ulna was readily accessible in all children, and the elbow could be placed at 90° or greater, enabling easy identification of the proximal end of the ulna. The ulna styloid process was readily identifiable in all children, and its identification was not limited by wrist contracture. Increasing difficulty was encountered in identifying the distal end of the medial malleolus and in obtaining a tibia length measurement as teno-equinovarus deformities of the ankle worsened in severity. Adequate positioning for forearm and lower leg length measurements was not possible for any child with wrist or ankle deformities, respectively.
In children, change in pulmonary function over time provides a sensitive measure to monitor the progress of respiratory disease. Predicted values of pulmonary function, rather than observed values are used to compensate for the expected increase in pulmonary function associated with growth (24). In an individual, flow and volume measurements track along percentile lines throughout childhood and adolescence (25). Accurate measurement of predictor variables is essential. Measurement of each of the limb segments is reproducible and provides an accurate alternative to height for prediction of pulmonary function. The precision of prediction equations for each of the anthropometric measurements was similar. Ulna length was chosen because it is readily accessible, even in wheelchair-bound children, and its measurement is unaffected by weakness or by joint or spinal deformity. In children with DMD, the elbow may not permit bending to 90°, nor the wrist full extension, but it was possible to adopt a reasonable position in all cases such that bony landmarks were readily identified and the measurement was unhindered. Forearm and lower leg measurements cross joints and could be affected by joint deformity such as contracture. Arm span measurement could also be affected by weakness or by joint or spinal deformity, and was included only for comparison. Tibia length is readily measured and unaffected by weakness, but palpating the distal end of the medial malleolus becomes particularly challenging in the face of severe equinovarus deformities of the ankle. In this study, prediction equations for pulmonary function using ulna length and age have similar precision (r2) to those using height. The degree of precision in prediction of FEV1 and FVC in this study is comparable with that of Zapletal and coworkers (14) (r2 = 0.86–0.92) and Hibbert and coworkers (21) (a study of Australian children; R2 = 0.85–0.92) who used height to predict pulmonary function. It is superior to that in the study of Knudsen and coworkers (11) (r2 = 0.53–0.80) and the Nhanes study (26) (r2 = 0.61–0.80). Spender and coworkers (27) compared measurements of upper arm length obtained using steel and plastic tape measures with those obtained using an anthrometer. The measurements obtained were on average 1.03 ± 0.20 cm and 1.10 ± 0.25 cm longer than the anthrometer measurement, respectively, and the intraobserver variability was greater (p = 0.002). A similar degree of inaccuracy would be expected when measuring the ulna with a tape measure and is not recommended. Ulna length measurement with a Harpenden caliper is simple, requiring only superficial palpation. The technique can be taught to new staff in minutes. Vernier calipers, that measure to 0.2 cm, are available from precision engineering stores for $A500 (Australian dollars) and may be used. Measurements were only performed on one side of the body as symmetry has previously been demonstrated (16, 28). The small values of the SD for individual values demonstrates the precision and reproducibility of the measurement. Several authors have previously documented the relationship between height and various body segments in their study populations (15, 16, 28–30). Particular attention has been paid to the upper limb (18, 28, 31) and arm span (17, 19, 31). In normal children, height prediction from upper limb measurements has high correlation coefficients (18). Arm span measurement is difficult and inaccurate in the presence of weakness or of joint or spinal deformity. It is currently being used by most respiratory laboratories to predict PFT in childhood neuromuscular weakness and spinal deformity due to the lack of a better alternative. Snyder and coworkers (32) have obtained anthropometric measurements of a large group of normal children in the United States. This normative data has been used to develop prediction equations for height from upper arm, forearm, ulna, lower leg, and arm span measurements (27, 28). Miller and Koreska (28) used the prediction equations for height from forearm, ulna, and arm span in three groups of children: normal children, children with idiopathic scoliosis, and children with DMD. In those with scoliosis, height was estimated by radiographic spine reconstruction. In children who are normal, have idiopathic scoliosis, or have DMD without wrist contractures, estimation of height from forearm measurements has a correlation coefficient of 0.96 (27). In children with DMD and wrist contractures, the ulna length to height correlation coefficient was 0.91. Interestingly, the arm span to height correlation coefficient is only 0.4748. Miller and Koreska (28) recommend that forearm is used if no wrist contractures exist, and ulna length is used if wrist contractures do exist. They recommend arm span not be used in children with DMD. In the current study, little difference was found between the precision of forearm or ulna measurements in predicting pulmonary function. Spender and coworkers (27) used prediction equations for height from upper arm and lower leg developed from Snyder and coworkers (32) data. Growth in cerebral palsy was assessed. It was found that height was reduced in children with spastic quadriplegia but was maintained in the normal range in children with diplegia and hemiplegia. There is evidence that limb growth may be impaired in upper motor neuron lesions, but the authors were unable to find evidence within the available literature that body proportions are altered in those with neuromuscular weakness or spinal deformity. Schools approached for this study consisted of public, private, religious, and independent primary and secondary schools. They were randomly selected from all metropolitan Melbourne schools to ensure adequate representation of socioeconomic and ethnic groups. Because of the multicultural nature of the Australian society (as represented by this sample), the prediction equations developed can be applicable in other multicultural societies with similar racial distribution. The response rate was lower than anticipated. This may be due, in part, to the request by many schools to give the students the responsibility for the questionnaire. When schools were agreeable, a recruitment package was sent directly to the parent or guardian, who returned the completed questionnaire and the consent form to the chief investigator in a reply-paid envelope. In developing prediction equations and normal reference ranges, all children with potential illnesses that may lead to them falling outside the normal reference range were excluded. This was clearly explained on the information sheet. Despite asking all children to return the form, even if this was the case, it may have led to a reduced return rate. The incidence of both asthma (33) and attention deficit hyperactivity disorder (34) is far greater in our community than suggested by the responses in this study. We believe participation bias to have little impact on the results, as this is not a prevalence study and only normal subjects were sought. In this study, pulmonary function measurements were performed with children in the standing position. In those with neuromuscular weakness, it is likely that these measurements will be performed in the sitting position. Both Townsend (35) and Lalloo and coworkers (36) have examined pulmonary function values obtained in the standing and sitting positions in adults. Lalloo and coworkers (36) have found the FEV1 in the standing position to be approximately 5% greater than that in the sitting position in women but less than 5% in men. Townsend (35) found the FEV1 and FVC to be 7 and 6% greater, respectively, in the standing position in men. The American Thoracic Society indicates that the VC is larger in the standing than in the sitting position in childhood, but this has not been quantitated (20). It is recommended that the same posture be adopted for each test. Normal PFT span a large range, and monitoring of progress is best achieved by comparing an individual's test with their previous results so that progress may be monitored. Height measurement is inaccurate in the presence of spinal deformity and when immobility is present. Spinal deformity causes restrictive pulmonary defects (37–39). Measurement of pulmonary function is particularly important in monitoring progress and in preoperative pulmonary risk assessment. Arm span measurements may be performed, but positioning for the measurement is often limited by the deformity of the spine, which may lead to an inequality of the level of the shoulders. This makes the measurement and the PFT predicted from it less accurate. In children with neuromuscular weakness, immobility is common, particularly in the later phases of the disease when the declining respiratory status is gaining importance. In DMD, the FVC rises in early childhood, plateaus in mid childhood, and then declines (6, 40). The age of acquisition and the magnitude of the plateau have prognostic implications (6). Children with neuromuscular weakness commonly have progressive spinal deformities that worsen the restrictive pulmonary defect and make major corrective spinal surgery necessary (1, 2, 41–43). Accurately predicted pulmonary function is important for monitoring disease progression, preoperative pulmonary risk assessment, guiding investigative pathways for sleep-disordered breathing, and instigation of respiratory supportive tools such as noninvasive ventilation and teaching techniques to assist mucociliary clearance (7, 44, 45).
Conclusions
The authors would like to acknowledge Ms. A. Stewart and Ms. K. Briggs for performing pulmonary function tests, Ms. Suzannah Vidmar for statistical computer assistance, and the students and staff of the participating schools.
Supported by Murdoch Children's Research Institute Royal Australasian College of Physicians. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: L.G.M. has no declared conflict of interest; J.K. has no declared conflict of interest; J.B.C. has no declared conflict of interest; C.F.R. has no declared conflict of interest. Received in original form March 28, 2003; accepted in final form July 14, 2003
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
