Methods for estimating body weight and height in hospitalized adults : a comparative analysis

The aim of this study was to compare the values obtained through methods directed to height and body weight estimates in relation to measurements taken from hospitalized adult. Study participants were 142 adults of both genders. Anthropometric measurements of body weight, height, knee height, arm length, span, demi-span, recumbent height, calf, arm and abdominal circumferences and subscapular skinfold thickness were taken. The actual measurements were compared with those obtained from formulas for estimating weight and height, using the paired t test. The estimated measurements differed significantly (p <0.001) from the actual measurements in both genders, observing the tendency of overestimating these measurements. The exception was the estimated height for men by the formula that utilizes the variable knee height (p> 0.001). The average estimated body weight closest to the actual body weight for men was obtained with the formula that used the measurements of arm, abdominal and calf circumferences. For women, the biggest coincidences were obtained by means of the formula that utilizes the variable knee height. For both men and women, the averages related to the body mass index, calculated through estimated body weight and height measurements resulted in the same nutritional diagnosis when compared to the body mass index involving actual measurements. The estimated height by the formula that utilizes the variable knee height among men was the only measurement which did not represent significant differences. Other methodologies for estimating body weight and height presented significant differences, which suggests that new studies using other methodologies are necessary.


INTRODUCTION
Nutritional assessment is a useful tool to support the therapeutic proposal and monitor the effectiveness of nutritional interventions, especially in hospitalized individuals 1 .Among the nutritional assessment methods, anthropometric measurements stand out, in which body weight and height are the most widely used.Both measures are essential for establishing the nutritional diagnosis and dietary and pharmacological prescriptions 2 .
Subject who are bedridden and unable to walk require equipment and technological solutions to meet the need for weighing in bed.Scales integrated to hospital beds are examples, however, they have high costs and are not reality in hospitals 3 .
Thus, many researchers have sought to develop methods to estimate body weight and height from specific measures of body segments that can be measured in these patients such as knee height, arm and calf circumferences, skinfold thickness, among others [4][5][6][7][8][9][10] .
Considering the importance of measures such as body weight and height as essential indicators in the assessment of the nutritional status and the impossibility to take these measures in bedridden subjects, this study aimed to compare values found by methods aimed to estimate body weight and height most frequently used in clinical practice with measurements taken in hospitalized adults.

METHODS
This is an analytical, quantitative cross-sectional study conducted in a public hospital in southern Brazil from July 2011 to August 2012.The sample was obtained by convenience and selected according to the following criteria: adults aged 20-60 years of both sexes, able to walk and hospitalized in medical and surgical clinics in the aforementioned hospital.Subjects with peripheral edema, ascites or anasarca, with limb amputation or paralysis or on dialysis were excluded.
The project was approved by the Ethics Committee on Human Research of the Federal University of Santa Catarina, Protocol 1107.All study participants signed the informed consent form.
Anthropometric measurements were selected according to variables contained in the formulas for estimating body weight and height selected in this study (Box 1).The final data collection was preceded by training of the researchers involved, seeking to standardize procedures and measurement techniques.
Measurements were taken in the morning after a fasting period of 10 hours and after urination.Anthropometric measurements taken were height, body weight, semi-span (SS) , span (S), recumbent height (RH), knee height (KH), arm length (AL), arm circumference (AC), calf circumference (CC), abdominal circumference (AC) and subscapular skinfold thickness (SEsf).Height was measured with Altura Exata® stadiometer fixed to a wall without footer, two meters high.To measure body weight, Tanita ® digital scales with maximum capacity of 150 kg was used with subjects standing barefoot and wearing only hospital garments.Both measures were assessed according to standards recommended by Jelliffe 11 .Body weight and height measurements were considered "gold standard" for comparisons performed in the present study.
To measure SS, the distance between the midpoint of the sternum and the distal phalanx of the right middle finger was measured by passing a flexible, inelastic measure tape parallel to the clavicle 4 .To assess S, measure tape was also used to measure the distance between the ends of the middle fingers of both hands with arms outstretched, leveling the shoulders 12 .
RH measurement involved the individual's length obtained with a flexible, inelastic measure tape after marking the stretcher with chalk through the triangle from top of the head to the foot sole with patient in the supine position and with bed in complete horizontal position 13 .
KH was measured with subject in the supine position with the right leg at an angle of ninety degrees with knee and ankle using Cescorf® caliper, comprising a fixed part, which was positioned in the plantar surface of the foot (heel) and a movable part pressed on the head of the patella (kneecap) 6 .
AL was measured in the right arm with forearm flexed at ninety degrees, and measure was taken from the tip of the acromion process of the scapula and the olecranon process of the ulna, with measure tape on the lateral side of the arm 14 .
Circumferences were measured with a flexible and inelastic measure tape on the right side of the body according to techniques proposed by Callaway et al. 15 .AC was measured at the midpoint between the acromion and the olecranon; CC in the maximum circumference of the calf muscle of the right leg and AC was assessed on the smallest horizontal circumference in the area between the ribs and the iliac crest.
SEsf was obtained by averaging three measurements with Lange® compass obliquely to the longitudinal axis.The caliper jaws applied 1 cm infero-lateral to the thumb and finger raising the fold, and the thickness is recorded to the rearest 0,1 cm.The orientation of the costal arches located one centimeter below the inferior angle of the right scapula 16 .
Nutritional status was evaluated through the body mass index (BMI), calculated using the Body Weight (kg) / height (m)² coefficient.Nutritional status was classified according to cutoffs recommended by the World Health Organization (WHO) 17 .Forty-seven BMI combinations were calculated using estimated measures.We chose to examine only those with more satisfactory results.
Data were analyzed using the STATISTICA software, version 7.0.All variables were tested for normality by the Kolmogorov-Smirnov test.Assuming that all variables considered were normally distributed, the paired t test was used.As level of statistical significance, p value ≤ 0.05 was adopted.

RESULTS
The study included 142 adults, 74 (52.1%) were male.The average age of the sample was 42.5 years ± 11.1 years.Regarding nutritional status, overweight was identified in 29% of women and 28% of men.
Tables 1 and 2 show statistical information regarding actual and estimated height and body weight measures according to the different methods adopted.; A 2 = WHO 5 ; A 3 e A 4 = Rabito et al. 2 ; A 5 = Gray et al. 13 ; A 6 = Chumlea et al. 7 ; A 7 = Cereda et al. 8.
With respect to height, it was found that the estimated measures differed significantly (p <0.001) from actual measures for men and women.The exception was height estimated by the formula of Chumlea et al. 7 (A6) for men, which showed no significant mean difference (p> 0.05) compared to the actual measure.This formula was also the only one who underestimated actual height for females.Formulas developed by Rabito et al. 2 (A3 and A4) showed the closest measures of actual height for females.
The formula that showed the largest average difference from the actual height for both sexes was the formula that used the semi-span measure (A1) 4 , overestimating the height by 9 cm for males and 6 cm for females.The formula of WHO (A2) 17 , which uses the wingspan measure, also showed significant difference from the actual measure, but with lower mean differences (3 cm for males and 5 cm for females).

DISCUSSION
All estimated measures significantly differed from actual measures in both sexes, except for height for men using the formula of Chumlea et al. 7 .
As in the present study, other authors 1,8,18 also found no significant differences when comparing actual height measurements to those estimated by Chumlea et al. 7 .This formula has the advantage of easy application, as it requires the measurement of a single measure, KH, requiring the use of one caliper.
One of the formulas of easy application and widely used in clinical practice to estimate height is the formula that used twice the half-span (A1) 4 .This measure showed the largest difference from the actual measured in the present study.Other studies 1,18 also found significant overestimation of the actual height with this measure.Other studies found different results [20][21][22][23] , obtaining statistical similarity between actual height and twice the half-spam.The formula developed by WHO 5 , which uses wingspan also resulted in statistically significant differences in this study, but with a smaller difference from the actual measurement.A similar result was found by Beghetto et al. 24 , who obtained a significant difference between actual measure and that estimated by WHO 5 , with a mean difference of 3 cm between them, which use was not recommended despite its easy application.
Recumbent height (A5) 13 also did not result in average height closer to the actual measure, overestimating 3 cm in height for males and 4 cm for females.A similar result was obtained by Rodrigues et al. 20 .
In relation to methods to estimate body weight, this study showed a significant difference between actual and estimated weight measures for both sexes.Similar results were obtained by other authors 2,18,19 .Dock-Nascimento et al. 23 and Sampaio et al. 25 performed a study with adults and the elderly and obtained significant similarity between actual weight and that estimated by Chumlea et al. 6 .
The best results of body weight estimation in this study were obtained using the formulas of Rabito et al. 2 for men and Chumlea et al. 6 and Rabito et al. 2 for women.The formula of Rabito et al. 2 was easy to be applied, which calculation uses only AL, AC and CC measured only with measure tape.The estimated body weight with the formula of Chumlea et al. 6 (P1) requires the SESF measurement, which may represent a negative factor because it requires measuring skinfold thickness, the researcher must be trained and the individual should remain seated during measurement.
With respect to BMI, it was identified that most measures showed significant difference from actual measures.For males, there was no significant difference when estimated and actual BMI were compared using actual body weight and that estimated by Chumlea et al. 7 (BMI1).This result was expected because this formula showed the best results to obtain height for men in the present study.BMI 5 and 8, which used weight (P2 and P3) and height (A4) measures estimated according to the formulas proposed by Rabito et al. 2 , also showed no difference compared to actual BMI values.It is believed that this result may be due to the similar body composition among males of both studies, as in this study, the mean BMI was 24.2 kg / m² and 24 kg / m² in the study of Rabito et al. 2 .
For females, there was no significant difference only between actual BMI and BMI 5 , which combined weight (P1) and height (A6) estimated by Chumlea et al. 7 .This result may be associated with the fact that these estimated measures are the closest to actual measures among females in this study.
In the present study, estimated BMI that showed results closer to actual values were those that combined weight measured using the best height estimate for men.Few studies that assessed BMI using estimated measures were found, which makes the comparison with our results difficult.

CONCLUSION
By comparing methodologies for estimating body weight and height used in this study with actual measures, it was observed that these methods showed a trend of overestimating body measures in both sexes.The only estimated measures that showed no significant difference from the actual measure was height for men, using anthropometric variable knee height.This measure is of easy application in clinical practice.
With respect to the body weight estimation, no results showed significant similarity.Whereas weight variation is a strong indicator to assess the effectiveness of the therapeutic procedures, methodological alternatives for obtaining this measure should be proposed.
Given that BMI is the most widely used nutritional diagnosis method, we sought to identify which combinations of estimated weight and height would result in BMI more close to actual BMI.It was observed that the use of body weight and height obtained from formulas developed by the same author resulted in BMI more close to actual BMI.
There are still few national studies comparing different methodologies for estimating body weight and height in adults and especially analyzing the nutritional status of populations from these measures.Thus, further research should be carried out in order to validate existing equations and compare specific population groups using formulas that combine effectiveness, low cost and easy application.

Box 1 .
Formulas analyzed in the present study.

Table 1 .
Comparison between actual and estimated height measures (m) of hospitalized adults SD = Standard deviation; A 1 = Mitchell & Lipschitz 4

Table 2 .
Comparison between actual and estimated body weight measures (kg) of hospitalized adults

Table 4 .
Comparison between actual and estimated body mass index values (kg/ m²) of hospitalized female adults.