Assessing the reliability of FTIR spectroscopy measurements and validity of bioelectrical impedance analysis as a surrogate measure of body composition among children and adolescents aged 8–19years attending schools in Kampala, Uganda

Assessing the reliability of FTIR spectroscopy measurements and validity of bioelectrical... Background: Accurate measurement of body composition in children and adolescents is important as the quantities of fat and fat-free mass have implications for health risk. The objectives of the present study were: to determine the reliability of Fourier Transform Infrared spectroscopy (FTIR) measurements and; compare the Fat Mass (FM), Fat Free Mass (FFM) and body fat percentage (%BF) values determined by bioelectrical impedance analysis (BIA) to those determined by deuterium dilution method (DDM) to identify correlations and agreement between the two methods. Methods: A cross-sectional study was conducted among 203 children and adolescents aged 8–19 years attending schools in Kampala city, Uganda. Pearson product-moment correlation at 5% significance level was considered for assessing correlations. Bland Altman analysis was used to examine the agreement between of FTIR measurements and between estimates by DDM and BIA.. Reliability of measurements was determined by Cronbach’salpha. Results: There was good agreement between the in vivo D O saliva enrichment measurements at 3 and 4 h among the studied age groups based on Bland-Altman plots. Cronbach’s alpha revealed that measurements of D Osaliva enrichment had very good reliability. For children and young adolescents, DDM and BIA gave similar estimates of FFM, FM, and %BF. Among older adolescents, BIA significantly over-estimated FFM and significantly under-estimated FM and %BF compared to estimates by DDM. The correlation between FFM, FM and %BF estimates by DDM and BIA was high and significant among young and older adolescents and for FFM among children. Conclusions: Reliability of the FTIR spectroscopy measurements was very good among the studied population. BIA is suitable for assessing body composition among children (8–9 years) and young adolescents (10–14 years) but not among older adolescents (15–19 years) in Uganda. The body composition measurements of older adolescents determined by DDM can be predicted using those provided by BIA using population-specific regression equations. Keywords: Body composition, Bioelectric impedance analysis, Deuterium dilution method, Children, Adolescents, agreement, reliability * Correspondence: catherinendagire@gmail.com School of Food Technology, Nutrition and Bio-engineering, Makerere University, Kampala, Uganda Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Ndagire et al. BMC Public Health (2018) 18:687 Page 2 of 9 Background the objectives of the present study were to: i) assess the Nutrition-related non-communicable diseases (NCDs) reliability of FTIR spectroscopy measurements of saliva such as hypertension, high blood glucose and cholesterol D O enrichment for determination of body composition levels, diabetes and cardiovascular diseases, are increasing and; ii) compare the body composition variables deter- and are predicted to be the major cause of morbidity and mined by BIA and DDM and, identify possible correla- mortality in most developing nations by 2020 [1]. In chil- tions and agreement between the two methods. dren and adolescents, the most common risk factors for nutrition-related NCDs include overweight, obesity, phys- Methods ical inactivity and unhealthy diets [2]. Pediatric and ado- Subjects lescent overweight and obesity are the driving force In a cross-sectional study, 203 apparently healthy (based behind metabolic syndrome risk that has become a grow- on self-proclamation) participants attending primary and ing public health concern in low and middle-income secondary schools in Kampala city, Uganda were se- countries (LMICs) [3]. This calls for interventions to pre- lected through a two-stage cluster sample design. The vent and manage childhood and adolescent overweight Ministry of Education and Sports provided an up to date and obesity. These interventions’ design, monitoring and list of all the primary and secondary schools in Kampala evaluation rely on correct identification of overweight and from which schools to participate in the study were ran- obese individuals. Therefore, there is need for accurate domly selected. Due to homogeneity between schools body composition measures to correctly identify over- and between students in divisions of Kampala, schools weight and obese individuals. were treated as clusters. Sampling of students from Body mass index (BMI) is the commonly used technique schools followed probability proportion to size proced- to determine nutrition status because it is inexpensive, fast ure and a sample of 203 participants aged 8–19 years and non-invasive [4]. However, it is a poor index of fatness was randomly selected using random numbers. and has poor sensitivity and inaccuracy for categorizing of Since sample size determination for validation studies obesity and overweight [5]. These limitations make BMI a is rarely ever justified a priori [17], for this study valid- poor outcome for research on the efficacy of nutrition pro- ation sample size was based on recommendations of re- grams. A number of reference methods are used to estimate searchers in the field of validity studies and from sample body composition, including underwater weighing (UWW) sizes used in previous validity studies as stated below. technique, air displacement plethysmography (ADP), For a study of agreement between two methods of meas- dual-energy X-ray absorptiometry (DEXA) and Deuterium urement, a sample size of 100 subjects is sufficient, giv- Dilution Method (DDM) [6]. WhileDDM haswidelybeen ing a 95% CI of about +/− 0.34 s, where s is the standard used due to its simplicity and relatively low cost, no pub- deviation of the differences between measurements by lished study was found on FTIR spectroscopy measure- the two methods. A sample of 200 subjects is better ments’ reliability among any population in Uganda. since it gives a 95% CI of about +/− 0.24 s [18]. A sample Reliability is defined as the degree of consistency and the size of 100 to 200 subjects is a reasonable size for valid- lack of error in a measurement [7]. Despite the scarcity of ation studies as it’s adequate for a range of likely degrees studies on the reliability of FTIR measurements, it is a pre- of validity and allows for appropriate deletion of some requisite for investigators aiming to validate a device or tech- subjects [19]. Furthermore, a minimum of 80 subjects nique to evaluate the reliability of the reference method, as for validity studies provides highly representative esti- lack of reliability often masks the actual effects and leads to mates of the main study samples [20]. For most studies, misinterpretation [8]. Internal reliability which measures re- sample sizes used have often been small, ranging from peatability of a tool is determined by Cronbach’salpha [7] 15 to 189 subjects [21–25]. Against this background, for and by Bland-Altman analysis [9]. this study, 203 participants were selected. At least four BIA is a rapid, cheap, safe and simple technique for subjects were targeted for each age. The subject to item measuring body composition both in the field and in ratio (n = 4) is the frequently recommended approach clinical settings [10], based on population-specific pre- when performing an exploratory factor analysis [17]. In dictive equations [11]. Since the validity of BIA measure- a similar study to assess body composition in Mexican ments varies with age and ethnicity [12], a number of school children of different geographical regions and studies have assessed the validity of BIA devices in vari- ethnicity, two children per age and ethnic group were ous populations of children and adolescents commonly regarded as sufficient [26]. using DEXA and DDM as reference techniques [13–16]. The selected subjects’ nutritional status was evaluated However, BIA’s validity for assessment of body compos- by anthropometric measurements: BMI, waist circumfer- ition and agreement with reference techniques like ence, waist to hip ratio and weight to height ratio and DDM has not been assessed among any population in their body composition was assessed by BIA and DDM. Uganda, including children and adolescents. Therefore, Immediately after the anthropometric and BIA Ndagire et al. BMC Public Health (2018) 18:687 Page 3 of 9 measurements were taken, saliva samples were collected instructions. The instrument was housed in the Depart- from the subjects and D O doses were given to them. ment of Biochemistry, Makerere University Kampala, This permitted the assessments to be performed at the Uganda. The instrument settings were: measurement same time and under the same conditions, with a conse- mode: absorbance; apodization: square triangle; number − 1 quent constant state of hydration during all methods of of scans: 32; resolution: 2.0 and; range (cm ): minimum body composition assessment used in the study. 2300 - maximum 2900. A ‘background’ scan was performed using the unen- Assessing height and weight riched drinking water that was used to make the calibra- Height and weight were taken by trained researchers tion standard (zero standard) and the instrument was using standard equipment. Body weight was measured calibrated using a prepared D O standard (1000 mg/kg). to the nearest 0.1 kg using a weighing scale, (Seca 899; Total body water (TBW) was calculated from the saliva Seca Weighing and Measuring Systems, Model No. sample by plateau method, based on the assumption that 8691321004, SECA Gmbh & Co. Germany made in this plateau was reached at 3 or 4 h. FM and %BF were China) with minimal clothing and no shoes. Height was estimated from TBW while FFM was calculated from measured to the nearest 0.1 cm using a height board FM. (Shorr-board, height board, Weight and Measure LLC, Irwin J. Shorr, MPH, MPS. Olney, Maryland USA) with- Statistical analysis out shoes. BMI (kg/m ) was calculated as weight in kilo- Descriptive statistics (means and confidence intervals) gram divided by the square of height in meters. were used for presentation of measurements data for D O enrichment, participants’ characteristics and body Assessing waist and hip circumferences composition (FFM, FM, and %BF) by DDM and BIA. Waist circumference (WC) was measured to the nearest Normality of variables was inspected visually using nor- 0.1 cm in standing position at the midpoint between the mal histogram plots. Box plots were used to inspect for lowest rib and the iliac crest and at the end of normal data outliers 8 of which were removed. To show the re- expiration, using a measuring tape. Hip circumference lationship between saliva D O enrichment at 3 and 4 h (HC) was measured to the nearest 0.1 cm in standing after ingestion of the D O dose when equilibration is position at the widest point of the hips using a measur- achieved, Pearson product-moment correlation was ing tape (Lufkin Executive Diameter Steel Tape, 2 m used. Reliability of the two FTIR measurements was Thinline Model W606 PM, Apex Tool Group, LLC NC verified using the Bland-Altman analysis by plotting the 27502, USA). differences between the two measurements of each sub- ject against the mean value of the two measurements. Body composition assessment by bioelectrical impedance Mean differences and limits of agreement were deter- analysis mined according to Bland Altman procedures. Limits of Body composition by BIA was measured using a BIA agreement were considered as the mean of differences (Tanita SC-331S Body Composition Analyzer; Tanita between the measurements at 3 and 4 h ± 1.96 × their Inc., Arlington Heights, IL) instrument, which provides standard deviation. Cronbach’s alpha was used to assess a measure of fat mass and fat-free mass using in-built the level of reliability of the FTIR spectroscopy measure- manufacturers’ equations. Impedance was measured ments at 3 and 4 h after D O dose ingestion. Cronbach’s with the subject standing barefoot on the metal α values between 0.7–0.9 were considered representative foot-plates of the machine for approximately 1 min. The of good reliability, while values above 0.9 were consid- subject’s age, gender, and height were entered into the ered representative of very good reliability [27]. machine, and a standard 0.5 kg was entered as an adjust- Paired t-tests were used to compare mean measures of ment for clothing weight for all participants. FFM, FM, and %BF by BIA and DDM. To show the rela- tionship between DDM and BIA, Pearson Body composition assessment using deuterium dilution product-moment correlation was considered. The Bland technique Altman plots examined the agreement between DDM A baseline saliva sample was collected from participants and BIA for measuring FFM, FM, and %BF. Mean differ- 2 hours after their last meal. Each participant then re- ences and limits of agreement were calculated according ceived an oral dose of D O (0.5 g/kg body weight). Two to Bland Altman procedures. Limits of agreement were endpoint saliva samples were collected at 3 and 4 h after considered as the mean of differences between measure- D O dose ingestion. Samples were stored in plastic saliva ments by DDM and BIA ± 1.96 × their standard devi- vials at − 20 °C until they were analyzed for D O using ation. The analyses were done using with STATA FTIR spectroscopy instrument (FTIR-8400S, Shimadzu version 13 software and the level of significance was set Corporation, Japan) according to manufacturer’s at P < 0.05. Ndagire et al. BMC Public Health (2018) 18:687 Page 4 of 9 Results among the different age groups (Fig. 1b). For children There were wide ranges for body weight, height, BMI, and young adolescents, FFM, FM and %BF estimates by waist circumference and hip circumference across the DDM were not statistically significantly different from different age groups (Table 1) In the current study, 16 those measured by BIA (Table 3). Among older adoles- children aged 8–9 years, 112 young adolescents aged cents, DDM significantly underestimated FFM (P < 10–14 years and 67 older adolescents aged 15–19 years; 0.0001) and significantly overestimated FM and %BF at 84 males and 111 females with mean (95% confidence P < 0.0001 and P < 0.0001 respectively compared to BIA. interval) age 13.44 (12.98 to 13.90) years, weight 44.61 Among young and older adolescents, the correlations (42.92 to 46.31) kg, height 1.51 (CI: 1.50, 1.53) m, waist between FFM, FM and % BF estimates by DDM and BIA circumference 65.87 (CI: 65.02, 66.71) cm and hip cir- were high and significant at r > 0.7 and P < 0.0001 cumference 82.48 (CI: 80.97, 83.99) cm participated. (Figs. 3a and 4a). The Bland-Altman plots for FFM, FM, Cronbach’s alpha values for the two measurements of and %BF showed a random nature of spread with no de- saliva D O enrichment were high (0.999, 0.997 and tectable significant negative bias for FFM, FM and % BF 0.996 for children, young and older adolescents, respect- values estimated by DDM and BIA among the different ively) (Table 2). age groups (Figs. 2b, 3b and 4b). The correlation coefficients for deuterium enrichment DDM and BIA exhibited generally narrower limits of at 3 and 4 h were high and positive among children, agreement for FFM, FM or % BF among children and young and older adolescents at r = 0.998, 0.995, and young adolescents than among older adolescents (Fig. 2b, 0.993 respectively (Fig. 1a). The Bland-Altman plots 3b, and 4b). Older adolescents (15–19-years) exhibited showed random nature of spread with no detectable pro- the largest mean differences for FFM (− 2.84 kg), FM portional bias for saliva D O enrichment at 3 and 4 h (2.84 kg), and %BF (5.01) while young adolescents (10– Table 1 Participants’ characteristics Mean (95% Confidence Interval) Characteristic Children (8–9 years) Young adolescents (10– Older adolescents (15– Overall 14 years) 19 years) N 16 112 67 195 Male 7 51 26 84 Female 9 61 41 111 Age 8.34 (8.11 to 8.64) 11.80 (11.57 to 12.04) 17.39 (17.09 to 17.69) 13.44 (12.98 to 13.90) Weight (kg) 28.31 (25.88 to 30.75) 40.40 (38.74 to 42.05) 55.55 (53.48 to 57.62) 44.61 (42.92 to 46.31) Height (m) 1.31 (1.28 to 1.35) 1.48 (1.47 to 1.50) 1.62 (1.60 to 1.64) 1.51 (1.50 to 1.53) BMI (kg/m ) 16.30 (15.65 to 16.95) 18.16 (17.69 to 18.63) 21.23 (20.53 to 21.93) 19.07 (18.64 to 19.50) Waist circumference (cm) 58.56 (56.83 to 60.28) 65.07 (64.06 to 66.08) 68.94 (67.62 to 70.26) 65.87 (65.02 to 66.71) Hip circumference (cm) 68.43 (66.25 to 70 .60) 79.20 (77.70 to 80.71) 91.32 (89.17 to 93.47) 82.48 (80.97 to 84.26) Waist height ratio 0.44 (0.44 to 0.46) 0.44 (0.43 to 0.44) 0.43 (0.42 to 0.44) 0.44 (0.43 to 0.44) Waist hip ratio 0.86 (0.84 to 0.88) 0.82 (0.82 to 0.83) 0.76 (0.74 to 0.78) 0.80 (0.80 to 0.81) 3 h deuterium enrichment 722.86 (626.83 to 796.77 (772.52 to 821.03) 790.85 (744.72 to 836.99) 788.67 (766.49 to (ppm) 818.88) 810.86) 4 h deuterium enrichment 720.06 (622.18 to 799.78 (775.74 to 823.83) 798.69 (753.26 to 844.11) 792.87 (770.84 to (ppm) 817.94) 814.89) DDM Total body water (litres) 17.18 (15.90 to 18.47) 24.97 (24.07 to 25. 87) 32.05 (30.75 to 33.35) 26.76 (25.84 to 27.68) DDM Total body water (%) 60.92 (59.38 to 62.47) 62.29 (61.44 to 63.14) 58.25 (56.21 to 60.28) 60.79 (59.90 to 61.68) DDM Fat free mass (kg) 23.48 (21.72 to 25.23) 34.11 (32.88 to 35.34) 43.78 (42.00 to 45.56) 36.56 (35.30 to 37.82) BIA Fat free mass (kg) 24.13 (22.10 to 26.16) 33.89 (32.72 to 35.06) 46.62 (45.05 to 48.19) 37.46 (36.13 to 38.79) DDM Fat mass (kg) 4.84 (3.90 to 5.77) 6.29 (5.61 to 6.96) 11.77 (9.97 to 13.58) 8.05 (7.23 to 8.87) BIA Fat mass (kg) 4.18 (3.48 to 4.88) 6.51 (5.80 to 7.22) 8.93 (7.54 to 10.32) 7.15 (6.50 to 7.81) DDM Fat (%) 16.77 (15.65 to 16.95) 14. 90 (13.74 to 16.06) 20.43 (17.64 to 23.71) 16.95 (15.74 to 18.17) BIA Fat (%) 14.61 (12.83 to 16.39) 15.34 (14.21 to 16.45) 15.42 (13.32 to 17.52) 15.30 (14.34 to 16.27) Impedance 660.00 (626.16 to 596.63 (581.35 to 611.90) 515.04 (500.91 to 529.16) 573.79 (561.67 to 693.84) 585.92) Ndagire et al. BMC Public Health (2018) 18:687 Page 5 of 9 Table 2 Cronbach’s alpha values of the two readings for the enrichment measurements were reproducible among the different age groups study population. The high Cronbach’s alpha value (> Age category Cronbach’s Alpha 0.9) among all studied age groups indicates very good re- peatability of the FTIR spectroscopy saliva D O enrich- 8–9 years 0.999 ment measurements among children, young and older 10–14 years 0.997 adolescents in Uganda. The FTIR spectroscopy instru- 15–19 years 0.996 ment can, therefore, provide reliable measures for D O saliva enrichment and thus suitable for validation of 14-years) exhibited lowest mean differences for FFM other body composition assessment techniques for more (0.22 kg), FM (− 0.22 kg) and %BF (− 0.44) (Table 3). accurate assessment of body composition among chil- Furthermore, the mean differences between DDM and dren and adolescents in Uganda. Furthermore, the FTIR BIA for measures of FFM, FM and %BF for older spectroscopy technique has several advantages in asses- adolescents exhibited largest 95% confidence intervals sing body composition including simplicity to carry out, compared to those for children and young adolescents. minimal subject cooperation requirements, acceptability The Bland Altman plots for FFM, FM, and %BF for older in all age groups [28], non-invasiveness, relatively low adolescents exhibited largest limits of agreement cost, easy administration of tracers, and easy collection compared those of children and young adolescents of samples [29]. (Fig. 2b, 3b and 4b). In this study, the ability of the inbuilt equations from the Tanita SC-331S BIA instrument to assess body com- Discussion position of children and adolescents in Uganda by using Prior to this work, no published studies were found on DDM as a reference method also was investigated. Prior the reliability of FTIR saliva D O enrichment measure- to this work, no published studies were found compar- ments among populations in Uganda. In this study, the ing the body composition estimates obtained by BIA to reliability of FTIR spectroscopy measurements among those obtained by DDM among children and adolescents children and adolescents in Uganda was assessed. The in Uganda. Since estimates for body composition had high and positive correlation between 3 and 4-h FTIR varying agreement across the studied age groups, DDM spectroscopy measurements is indicative of similarity and BIA are generally not interchangeable across chil- and reproducibility of the two sets of measurements. dren and adolescents in Uganda. The none-statistically The Bland-Altman plots that showed no apparent trend significantly different (P > 0.05) FFM, FM and %BF mea- in error differences between the measurements taken sures by DDM and BIA among children and young ado- after 3 and those taken after 4 h imply that saliva D O lescents imply possibility for agreement between the two Fig. 1 Regression and Bland-Altman plots for saliva D O enrichment among children (left), young adolescents (middle) and older adolescents (right) Ndagire et al. BMC Public Health (2018) 18:687 Page 6 of 9 Table 3 Body composition mean values (CI), mean difference (CI) and P-values between DDM and BIA among different age groups Body composition means DDM BIA Mean difference P-value Children (8–9 years) FFM (kg) 23.48 (21.72–25.23) 24.13 (22.10–26.16) −0.657 (−1.474–0.160) 0.1071 FM (kg) 4.84 (3.90–5.77) 4.18 (3.48–4.88) 0.657 (− 0.160–1.474) 0.1071 %BF 16.77 (14.66–18.88) 14.61 (12.83–16.39) 2.157 (−0.397–4.711) 0.0920 Young adolescents (10–14 years) FFM (kg) 34.11 (32.88–35.34) 33.89 (32.72–35.06) 0.224 (− 0.111–0.559) 0.1876 FM (kg) 6.29 (5.61–6.96) 6.51 (5.80–7.22) −0.224 (− 0.559–0.111) 0.1876 %BF 14. 90 (13.74–16.06) 15.34 (14.21–16.45) −0.436 (− 1.239–0.367) 0.2846 Older adolescents (15–19 years) FFM (kg) 43.78 (42.00–45.56) 46.62 (45.05–48.19) −2.841 (− 3.983 - -1.699) < 0.0001 FM (kg) 11.77 (9.97–13.58) 8.93 (7.54–10.32) 2.841 (1.699–3.983) < 0.0001 %BF 20.43 (17.64–23.71) 15.42 (13.32–17.52) 5.006 (3.068–6.944) < 0.0001 methods in these age categories. For children and young validate 2 portable BIA devices; the Inbody 230 and the adolescents, the generally narrow limits of agreement, Tanita BC-418 for body composition assessment in the small mean discrepancies (biases) for the FFM, FM healthy Taiwanese school-age children, Bland-Altman and %BF estimates and their narrow 95% confidence in- analysis showed clinically acceptable agreement between tervals of means imply that DDM and BIA estimates for the Inbody 230 device and DEXA for FFM measure- FFM, FM, and %BF agree and can be used interchange- ments [15]. ably for either FM, FFM, or %BF for these age categories On the other hand, the statistically significantly differ- in Uganda. These findings are similar to those by Mehta ent mean values (P < 0.05) for FFM, FM and %BF among and others who found agreement between BIA and older adolescents imply no possibility for agreement be- DDM for FFM, FM and %BF among children 14 years of tween the two methods. The wide limits of agreement age or younger with Intestinal Failure [23]. In a study to for FFM, FM, or %BF exhibited by Bland Altman plots Fig. 2 Regression and Bland-Altman plots for FFM (left), FM (middle) and % body fat (right) determined by DDM and BIA among children Ndagire et al. BMC Public Health (2018) 18:687 Page 7 of 9 Fig. 3 Regression and Bland-Altman plots for FFM (left), FM (middle) and % body fat (right) determined by DDM and BIA among young adolescents for older adolescents, the big mean discrepancies interchangeable for either FM, FFM, or %BF among (biases) for the FFM, FM and %BF estimates and their older adolescents (15–19 years) in Uganda. Similar to wide 95% confidence intervals in this age group imply this study’s findings where BIA overestimated FFM limited agreement between the two methods. This re- among older adolescent and underestimated their FM veals that DDM and BIA are not directly and %BF are those by Resende and others who reported Fig. 4 Regression and Bland-Altman plots for FFM (left), FM (middle) and % body fat (right) determined by DDM and BIA among old adolescents Ndagire et al. BMC Public Health (2018) 18:687 Page 8 of 9 that BIA overestimated the measures of FFM and under- composition in resource-poor countries that cannot af- estimated the measures of FM compared to those pro- ford four-compartment (gold standard) techniques. vided by DDM among obese adolescents in Brazil [5]. The other results of the study showed that DDM and Resende and others reported a high, positive and signifi- BIA can be used interchangeably for FFM, FM, and %BF cant correlation between FFM and FM values deter- for children and young adolescents aged 8–14 years in mined by DDM and BIA but there was no agreement Uganda but not interchangeable for the assessment of between the two methods among obese adolescents [5] body composition in older adolescents aged 15–19 years as was the case for older adolescents in this study. In a in Uganda. For that reason, among older adolescents in study to validate predictive equations of BIA to FFM es- Uganda, BIA is not a valid measure for body compos- timation in army cadets aged 17–24 years, Langer and ition, so deriving population-specific BIA equations may others observed significant differences between FFM be a suitable approach for assessing body composition. values from 8 predictive BIA equations and no good The study, therefore, revealed BIA’s limitations in asses- agreement with DXA [11] Also, among healthy Indian sing body composition among children and adolescents children and adolescents aged 5–18 years, there was no in Uganda. agreement between BIA and DXA in assessment of body Abbreviations composition [13]. A possible explanation for the discrep- 2 %BF: Body fat percentage; H: Deuterium; ADP: Air displacement ancy of body composition among older adolescents by plethysmography; BIA: Bioelectrical impedance analysis; BMI : Body mass index; D O: Deuterium oxide; DDM: Deuterium dilution method; DEXA: Dual- the BIA system’s inbuilt prediction equations is that they energy x-ray absorptiometry; FFM: Fat-free mass; FM: Fat mass; FTIR: Fourier are normally based on Western European or North transform infrared; IAEA: International atomic energy agency; LMICs: Low and American populations, which may differ in body com- middleiIncome countries; NCDs : Non-communicable diseases; TBW: Total body water; UWW : Underwater weighing; VD: Dilution space; WHO: World position and proportion when compared to the popula- Health Organization tion under study [30]. Growth involves the deposition of both fat mass (FM) and fat-free mass (FFM) components Acknowledgements The authors are grateful to the children and adolescents who participated in and human body composition is ethnicity dependent this study, their parents/guardians, schools’ administration and teachers and [31]. No literature was found regarding the age- and the entire research team. sex-related pattern of changes in body composition for Funding populations in Uganda. The International Atomic Energy Agency (IAEA) offered the study support While the study was the first of its kind among under project number UGA6017. populations in Uganda, it was not without limitations. DDM was used as the reference method which, Availability of data and materials The datasets used and/or analyzed during the current study are available although widely validated as a reliable estimate, is not from the corresponding author on reasonable request. a gold standard for body composition. Ideally, a four-component model would have been used as the ref- Authors’ contributions CTN, DI, JHM, JEA, and DN conceived, designed, and revised the manuscript. erence method, but this was not possible in our study CTN did the literature search. SMAS, RB, and BO did the statistical analysis. All setting. While DDM has the advantage that it is rela- authors read and approved the final manuscript. tively easy to perform, it is not without limitations: one Ethics approval and consent to participate assumption is the hydration of FFM, which may vary The purpose and objectives of the study were carefully explained to each among persons by age, sex, maturation and ethnicity participant and their parents. Informed consent to the study was obtained and to estimate FFM from TBW, age, and sex-specific from participants’ parent/guardian to affirm their willingness or not. The parents or guardians of participants provided consent to allow their children hydration fractions were used [6]. But the hydration of to take part in the study while participants signed assent accepting to FFM values used for computation of TBW to estimate participate in the study. Ethical clearance to engage human subjects was FFM, FM, and %BF were not Uganda specific. Higher obtained from Makerere University School of Biomedical Sciences Higher Degrees, Research and Ethics Committee and Uganda National Council for hydration factors have been observed among African Science and Technology under reference numbers: SBS 291 and HS 1950 American adults compared to whites using a respectively. four-component model [32]. However, there is no infor- Competing interests mation on the hydration factors of FFM for Ugandan The authors declare that they have no competing interests. populations. Publisher’sNote Conclusions Springer Nature remains neutral with regard to jurisdictional claims in The reliability of the FTIR spectroscopy saliva D O en- published maps and institutional affiliations. richment measurements was very good among the stud- Author details ied population. This technique can be used as a School of Food Technology, Nutrition and Bio-engineering, Makerere reference technique in the validation of field techniques 2 University, Kampala, Uganda. Department of Biochemistry and Sports like BIA for more accurate estimation of body Science, Makerere University, Kampala, Uganda. Department of Statistical Ndagire et al. BMC Public Health (2018) 18:687 Page 9 of 9 Methods and Actuarial Science, School of Statistics, Makerere University, stud 2015. https://www-users.york.ac.uk/~mb55/meas/sizemeth.htm. 4 5 Kampala, Uganda. Centre MURAZ, Bobo-Dioulasso, Burkina Faso. Division of Accessed 7 Feb 2018. Nutritional Sciences, University of Illinois, Urbana-Champaign, USA. 19. Willet W. Nutritional epidemiology. Oxford: Oxford University Press; 1998. Department of Food Science and Human Nutrition, University of Illinois, 20. Hobart JC, Cano SJ, Warner TT, Thompson AJ. What sample sizes for Urbana-Champaign, USA. reliability and validity studies in neurology? J Neurol. 2012;259:2681–94. https://doi.org/10.1007/s00415-012-6570-y. Received: 3 November 2017 Accepted: 29 May 2018 21. Bosy-Westphal A, Later W, Hitze B, Sato T, Kossel E, Glüer C-C, et al. Accuracy of bioelectrical impedance consumer devices for measurement of body composition in comparison to whole body magnetic resonance imaging and dual X-ray absorptiometry. Obes Facts. 2008;1:319–24. https://doi.org/ References 10.1159/000176061. 1. Celermajer DS, Chow CK, Marijon E, Anstey NM, Woo KS. Cardiovascular 22. Li YC, Li CI, Lin WY, Liu CS, Hsu HS, Lee CC, et al. Percentage of body fat disease in the developing world. J Am Coll Cardiol. 2012;60:1207–16. assessment using bioelectrical impedance analysis and dual-energy X-ray https://doi.org/10.1016/j.jacc.2012.03.074. absorptiometry in a weight loss program for obese or overweight Chinese 2. Raj M. Obesity and cardiovascular risk in children and adolescents. Indian J adults. PLoS One. 2013;8:2–8. https://doi.org/10.1371/journal.pone.0058272. Endocrinol Metab. 2012;16:13. https://doi.org/10.4103/2230-8210.91176. 23. Mehta NM, Raphael B, Guteirrez IM, Quinn N, Mitchell PD, Litman HJ, et al. 3. WHO. Obesity and overweight @ WwwWhoInt 2017. http://www.who.int/ Comparison of body composition assessment methods in pediatric es/news-room/fact-sheets/detail/obesity-and-overweight. Accessed 1 June intestinal failure. J Pediatr Gastroenterol Nutr. 2014;59:99–105. https://doi. org/10.1097/MPG.0000000000000364. 4. Akindele MO, Phillips JS, Igumbor EU. The relationship between body fat 24. Shafer KJ, Siders WA, Johnson LK, Lukaski HC. Validity of segmental multiple- percentage and body mass index in overweight and obese individuals in an frequency bioelectrical impedance analysis to estimate body composition of urban african setting. J Public Health Africa. 2016;7. https://doi.org/10.4081/ adults across a range of body mass indexes. Nutrition. 2009;25:25–32. jphia.2016.515. https://doi.org/10.1016/j.nut.2008.07.004. 5. Resende CMM, Camelo Júnior JS, Vieira MNCM, Ferriolli E, Pfrimer K, 25. Widen EM, Strain G, King WC, Yu W, Lin S, Goodpaster B, et al. Validity of Perdoná GSC, et al. Body composition measures of obese adolescents by bioelectrical impedance analysis for measuring changes in body water and the deuterium oxide dilution method and by bioelectrical impedance. percent fat after bariatric surgery. Obes Surg. 2014;24:847–54. https://doi. Brazilian J Med Biol Res. 2011;44:1164–70. https://doi.org/10.1590/S0100- org/10.1007/s11695-014-1182-5. 879X2011007500122. 26. Ramírez E, Valencia ME, Bourges H, Espinosa T, Moya-Camarena SY, Salazar 6. Srichan W, Yamborisut U, Kijboonchoo K, Thasanasuwan WA. G, et al. Body composition prediction equations based on deuterium oxide Comparison of bioelectrical impedance analysis with deuterium dilution dilution method in Mexican children: a national study. Eur J Clin Nutr. 2012; technique for body fat assessment in school-age children. J Public 66:1099–103. https://doi.org/10.1038/ejcn.2012.89. Health (Bangkok). 2014;44:223–36. 27. Chirita-Emandi A, Dobrescu A, Papa M, Puiu M, Emandi AC. Reliability of 7. Molina KM, Molina KM, Goltz HH, Kowalkouski MA, Hart SL, Latini D, et al. measuring Subcutaneous fat Tissue thickness using Ultrasound in non- Reliability and validity, Encycl. Behav. Med. New York, NY: Springer New athletic young adults MEASURING SUBCUTANEOUS FAT TISSUE USING York; 2013. p. 1643–4. https://doi.org/10.1007/978-1-4419-1005-9_1549. ULTRASOUND. Maedica -a J Clin med Maedica a J Clin med J Clin med 8. Meissl K, Smidt E, Tintner J. Reproducibility of fourier transform infrared Maedica a. J Clin Med. 2015;10:204–9. spectra of compost, municipal solid waste, and landfill material. Appl 28. IAEA. Nutritional and Health-Related Environmental Studies (NAHRES). Using Spectrosc. 2008;62:190–6. https://doi.org/10.1366/000370208783575537. nuclear techniques to assess the role of nutrition-sensitive agri-food systems 9. Bland MJ, Altman DG. Statistical methods for assessing agreement between in improving diet, health and nutritional status of vulnerable populations. two methods of clinical measurement. Lancet London J Onwhyn, 1823- 2015. http://www-naweb.iaea.org/nahu/NAHRES/crp/e43029.html. Accessed Oxford Elsevier 1986;327:307–10. doi:https://doi.org/10.1016/S0140- 14 Feb 2018. 6736(86)90837-8. 29. Lee SY, Gallagher D. Assessment methods in human body composition. 10. Böhm A, Heitmann BL. The use of bioelectrical impedance analysis for body Curr Opin Clin Nutr Metab Care. 2008;11:566–72. https://doi.org/10.1097/ composition in epidemiological studies. Eur J Clin Nutr. 2013;67:S79–85. MCO.0b013e32830b5f23.Assessment. https://doi.org/10.1038/ejcn.2012.168. 30. Wickramasinghe V, Lamabadusuriya S, Cleghorn G, Davies P. Assesment of 11. Langer RD, Borges JH, Pascoa MA, Cirolini VX, Guerra-Júnior G, Gonçalves body composition in Sri Lankachildren: validation of a biolectrical EM. Validity of bioelectrical impedance analysis to estimation fat-free mass impedance prediction equation. Eur J Clin Nutr. 2008;62:1170–7. https://doi. in the army cadets. Nutrients. 2016;8 https://doi.org/10.3390/nu8030121. org/10.1016/0377-8401(90)90166-6. 12. Dehghan M, Merchant AT. Is bioelectrical impedance accurate for use in 31. Xiong K-Y, He H, Zhang Y-M, Ni G-X. Analyses of body composition large epidemiological studies? Nutr J. 2008;7:26. https://doi.org/10.1186/ charts among younger and older Chinese children and adolescents 1475-2891-7-26. aged 5 to 18 years. BMC Public Health. 2012;12:835. https://doi.org/10. 13. Chiplonkar S, Kajale N, Ekbote V, Mandlik R, Parthasarathy L, Khadilkar V, et 1186/1471-2458-12-835. al. Validation of bioelectric impedance analysis against dual-energy X-ray 32. Prins M, Hawkesworth S, Wright A, Fulford a JC, LM a J, Prentice a M, et al. absorptiometry for assessment of body composition in Indian children aged Use of bioelectrical impedance analysis to assess body composition in rural 5 to 18 years. Indian Pediatr. 2017;54:919–24. Gambian children. Eur J Clin Nutr. 2008;62:1065–74. https://doi.org/10.1038/ 14. Haroun D, Croker H, Viner RM, Williams JE, Darch TS, Fewtrell MS, et al. sj.ejcn.1602830. Validation of BIA in obese children and adolescents and re-evaluation in a longitudinal study. Obesity. 2009;17:2245–50. https://doi.org/10. 1038/oby.2009.98. 15. Lee L-W, Liao Y-S, Lu H-K, Hsiao P-L, Chen Y-Y, Chi C-C, et al. Validation of two portable bioelectrical impedance analyses for the assessment of body composition in school age children. PLoS One. 2017;12:e0171568. https:// doi.org/10.1371/journal.pone.0171568. 16. Wang L, Hui SS. Validity of four commercial bioelectrical impedance scales in measuring body fat among Chinese children and adolescents. Biomed Res Int. 2015;2015:1–8. https://doi.org/10.1155/2015/614858. 17. Anthoine E, Moret L, Regnault A, Sébille V, Hardouin J-B. Sample size used to validate a scale: a review of publications on newly-developed patient reported outcomes measures. Health Qual Life Outcomes. 2014;12:2. https:// doi.org/10.1186/s12955-014-0176-2. 18. Bland M. How can I decide the sample size for a study of agreement between two methods of measurement? Freq asked quest des anal Meas http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Public Health Springer Journals

Assessing the reliability of FTIR spectroscopy measurements and validity of bioelectrical impedance analysis as a surrogate measure of body composition among children and adolescents aged 8–19years attending schools in Kampala, Uganda

Free
9 pages

Loading next page...
 
/lp/springer_journal/assessing-the-reliability-of-ftir-spectroscopy-measurements-and-MWZTsx0LG7
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s).
Subject
Medicine & Public Health; Public Health; Medicine/Public Health, general; Epidemiology; Environmental Health; Biostatistics; Vaccine
eISSN
1471-2458
D.O.I.
10.1186/s12889-018-5627-y
Publisher site
See Article on Publisher Site

Abstract

Background: Accurate measurement of body composition in children and adolescents is important as the quantities of fat and fat-free mass have implications for health risk. The objectives of the present study were: to determine the reliability of Fourier Transform Infrared spectroscopy (FTIR) measurements and; compare the Fat Mass (FM), Fat Free Mass (FFM) and body fat percentage (%BF) values determined by bioelectrical impedance analysis (BIA) to those determined by deuterium dilution method (DDM) to identify correlations and agreement between the two methods. Methods: A cross-sectional study was conducted among 203 children and adolescents aged 8–19 years attending schools in Kampala city, Uganda. Pearson product-moment correlation at 5% significance level was considered for assessing correlations. Bland Altman analysis was used to examine the agreement between of FTIR measurements and between estimates by DDM and BIA.. Reliability of measurements was determined by Cronbach’salpha. Results: There was good agreement between the in vivo D O saliva enrichment measurements at 3 and 4 h among the studied age groups based on Bland-Altman plots. Cronbach’s alpha revealed that measurements of D Osaliva enrichment had very good reliability. For children and young adolescents, DDM and BIA gave similar estimates of FFM, FM, and %BF. Among older adolescents, BIA significantly over-estimated FFM and significantly under-estimated FM and %BF compared to estimates by DDM. The correlation between FFM, FM and %BF estimates by DDM and BIA was high and significant among young and older adolescents and for FFM among children. Conclusions: Reliability of the FTIR spectroscopy measurements was very good among the studied population. BIA is suitable for assessing body composition among children (8–9 years) and young adolescents (10–14 years) but not among older adolescents (15–19 years) in Uganda. The body composition measurements of older adolescents determined by DDM can be predicted using those provided by BIA using population-specific regression equations. Keywords: Body composition, Bioelectric impedance analysis, Deuterium dilution method, Children, Adolescents, agreement, reliability * Correspondence: catherinendagire@gmail.com School of Food Technology, Nutrition and Bio-engineering, Makerere University, Kampala, Uganda Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Ndagire et al. BMC Public Health (2018) 18:687 Page 2 of 9 Background the objectives of the present study were to: i) assess the Nutrition-related non-communicable diseases (NCDs) reliability of FTIR spectroscopy measurements of saliva such as hypertension, high blood glucose and cholesterol D O enrichment for determination of body composition levels, diabetes and cardiovascular diseases, are increasing and; ii) compare the body composition variables deter- and are predicted to be the major cause of morbidity and mined by BIA and DDM and, identify possible correla- mortality in most developing nations by 2020 [1]. In chil- tions and agreement between the two methods. dren and adolescents, the most common risk factors for nutrition-related NCDs include overweight, obesity, phys- Methods ical inactivity and unhealthy diets [2]. Pediatric and ado- Subjects lescent overweight and obesity are the driving force In a cross-sectional study, 203 apparently healthy (based behind metabolic syndrome risk that has become a grow- on self-proclamation) participants attending primary and ing public health concern in low and middle-income secondary schools in Kampala city, Uganda were se- countries (LMICs) [3]. This calls for interventions to pre- lected through a two-stage cluster sample design. The vent and manage childhood and adolescent overweight Ministry of Education and Sports provided an up to date and obesity. These interventions’ design, monitoring and list of all the primary and secondary schools in Kampala evaluation rely on correct identification of overweight and from which schools to participate in the study were ran- obese individuals. Therefore, there is need for accurate domly selected. Due to homogeneity between schools body composition measures to correctly identify over- and between students in divisions of Kampala, schools weight and obese individuals. were treated as clusters. Sampling of students from Body mass index (BMI) is the commonly used technique schools followed probability proportion to size proced- to determine nutrition status because it is inexpensive, fast ure and a sample of 203 participants aged 8–19 years and non-invasive [4]. However, it is a poor index of fatness was randomly selected using random numbers. and has poor sensitivity and inaccuracy for categorizing of Since sample size determination for validation studies obesity and overweight [5]. These limitations make BMI a is rarely ever justified a priori [17], for this study valid- poor outcome for research on the efficacy of nutrition pro- ation sample size was based on recommendations of re- grams. A number of reference methods are used to estimate searchers in the field of validity studies and from sample body composition, including underwater weighing (UWW) sizes used in previous validity studies as stated below. technique, air displacement plethysmography (ADP), For a study of agreement between two methods of meas- dual-energy X-ray absorptiometry (DEXA) and Deuterium urement, a sample size of 100 subjects is sufficient, giv- Dilution Method (DDM) [6]. WhileDDM haswidelybeen ing a 95% CI of about +/− 0.34 s, where s is the standard used due to its simplicity and relatively low cost, no pub- deviation of the differences between measurements by lished study was found on FTIR spectroscopy measure- the two methods. A sample of 200 subjects is better ments’ reliability among any population in Uganda. since it gives a 95% CI of about +/− 0.24 s [18]. A sample Reliability is defined as the degree of consistency and the size of 100 to 200 subjects is a reasonable size for valid- lack of error in a measurement [7]. Despite the scarcity of ation studies as it’s adequate for a range of likely degrees studies on the reliability of FTIR measurements, it is a pre- of validity and allows for appropriate deletion of some requisite for investigators aiming to validate a device or tech- subjects [19]. Furthermore, a minimum of 80 subjects nique to evaluate the reliability of the reference method, as for validity studies provides highly representative esti- lack of reliability often masks the actual effects and leads to mates of the main study samples [20]. For most studies, misinterpretation [8]. Internal reliability which measures re- sample sizes used have often been small, ranging from peatability of a tool is determined by Cronbach’salpha [7] 15 to 189 subjects [21–25]. Against this background, for and by Bland-Altman analysis [9]. this study, 203 participants were selected. At least four BIA is a rapid, cheap, safe and simple technique for subjects were targeted for each age. The subject to item measuring body composition both in the field and in ratio (n = 4) is the frequently recommended approach clinical settings [10], based on population-specific pre- when performing an exploratory factor analysis [17]. In dictive equations [11]. Since the validity of BIA measure- a similar study to assess body composition in Mexican ments varies with age and ethnicity [12], a number of school children of different geographical regions and studies have assessed the validity of BIA devices in vari- ethnicity, two children per age and ethnic group were ous populations of children and adolescents commonly regarded as sufficient [26]. using DEXA and DDM as reference techniques [13–16]. The selected subjects’ nutritional status was evaluated However, BIA’s validity for assessment of body compos- by anthropometric measurements: BMI, waist circumfer- ition and agreement with reference techniques like ence, waist to hip ratio and weight to height ratio and DDM has not been assessed among any population in their body composition was assessed by BIA and DDM. Uganda, including children and adolescents. Therefore, Immediately after the anthropometric and BIA Ndagire et al. BMC Public Health (2018) 18:687 Page 3 of 9 measurements were taken, saliva samples were collected instructions. The instrument was housed in the Depart- from the subjects and D O doses were given to them. ment of Biochemistry, Makerere University Kampala, This permitted the assessments to be performed at the Uganda. The instrument settings were: measurement same time and under the same conditions, with a conse- mode: absorbance; apodization: square triangle; number − 1 quent constant state of hydration during all methods of of scans: 32; resolution: 2.0 and; range (cm ): minimum body composition assessment used in the study. 2300 - maximum 2900. A ‘background’ scan was performed using the unen- Assessing height and weight riched drinking water that was used to make the calibra- Height and weight were taken by trained researchers tion standard (zero standard) and the instrument was using standard equipment. Body weight was measured calibrated using a prepared D O standard (1000 mg/kg). to the nearest 0.1 kg using a weighing scale, (Seca 899; Total body water (TBW) was calculated from the saliva Seca Weighing and Measuring Systems, Model No. sample by plateau method, based on the assumption that 8691321004, SECA Gmbh & Co. Germany made in this plateau was reached at 3 or 4 h. FM and %BF were China) with minimal clothing and no shoes. Height was estimated from TBW while FFM was calculated from measured to the nearest 0.1 cm using a height board FM. (Shorr-board, height board, Weight and Measure LLC, Irwin J. Shorr, MPH, MPS. Olney, Maryland USA) with- Statistical analysis out shoes. BMI (kg/m ) was calculated as weight in kilo- Descriptive statistics (means and confidence intervals) gram divided by the square of height in meters. were used for presentation of measurements data for D O enrichment, participants’ characteristics and body Assessing waist and hip circumferences composition (FFM, FM, and %BF) by DDM and BIA. Waist circumference (WC) was measured to the nearest Normality of variables was inspected visually using nor- 0.1 cm in standing position at the midpoint between the mal histogram plots. Box plots were used to inspect for lowest rib and the iliac crest and at the end of normal data outliers 8 of which were removed. To show the re- expiration, using a measuring tape. Hip circumference lationship between saliva D O enrichment at 3 and 4 h (HC) was measured to the nearest 0.1 cm in standing after ingestion of the D O dose when equilibration is position at the widest point of the hips using a measur- achieved, Pearson product-moment correlation was ing tape (Lufkin Executive Diameter Steel Tape, 2 m used. Reliability of the two FTIR measurements was Thinline Model W606 PM, Apex Tool Group, LLC NC verified using the Bland-Altman analysis by plotting the 27502, USA). differences between the two measurements of each sub- ject against the mean value of the two measurements. Body composition assessment by bioelectrical impedance Mean differences and limits of agreement were deter- analysis mined according to Bland Altman procedures. Limits of Body composition by BIA was measured using a BIA agreement were considered as the mean of differences (Tanita SC-331S Body Composition Analyzer; Tanita between the measurements at 3 and 4 h ± 1.96 × their Inc., Arlington Heights, IL) instrument, which provides standard deviation. Cronbach’s alpha was used to assess a measure of fat mass and fat-free mass using in-built the level of reliability of the FTIR spectroscopy measure- manufacturers’ equations. Impedance was measured ments at 3 and 4 h after D O dose ingestion. Cronbach’s with the subject standing barefoot on the metal α values between 0.7–0.9 were considered representative foot-plates of the machine for approximately 1 min. The of good reliability, while values above 0.9 were consid- subject’s age, gender, and height were entered into the ered representative of very good reliability [27]. machine, and a standard 0.5 kg was entered as an adjust- Paired t-tests were used to compare mean measures of ment for clothing weight for all participants. FFM, FM, and %BF by BIA and DDM. To show the rela- tionship between DDM and BIA, Pearson Body composition assessment using deuterium dilution product-moment correlation was considered. The Bland technique Altman plots examined the agreement between DDM A baseline saliva sample was collected from participants and BIA for measuring FFM, FM, and %BF. Mean differ- 2 hours after their last meal. Each participant then re- ences and limits of agreement were calculated according ceived an oral dose of D O (0.5 g/kg body weight). Two to Bland Altman procedures. Limits of agreement were endpoint saliva samples were collected at 3 and 4 h after considered as the mean of differences between measure- D O dose ingestion. Samples were stored in plastic saliva ments by DDM and BIA ± 1.96 × their standard devi- vials at − 20 °C until they were analyzed for D O using ation. The analyses were done using with STATA FTIR spectroscopy instrument (FTIR-8400S, Shimadzu version 13 software and the level of significance was set Corporation, Japan) according to manufacturer’s at P < 0.05. Ndagire et al. BMC Public Health (2018) 18:687 Page 4 of 9 Results among the different age groups (Fig. 1b). For children There were wide ranges for body weight, height, BMI, and young adolescents, FFM, FM and %BF estimates by waist circumference and hip circumference across the DDM were not statistically significantly different from different age groups (Table 1) In the current study, 16 those measured by BIA (Table 3). Among older adoles- children aged 8–9 years, 112 young adolescents aged cents, DDM significantly underestimated FFM (P < 10–14 years and 67 older adolescents aged 15–19 years; 0.0001) and significantly overestimated FM and %BF at 84 males and 111 females with mean (95% confidence P < 0.0001 and P < 0.0001 respectively compared to BIA. interval) age 13.44 (12.98 to 13.90) years, weight 44.61 Among young and older adolescents, the correlations (42.92 to 46.31) kg, height 1.51 (CI: 1.50, 1.53) m, waist between FFM, FM and % BF estimates by DDM and BIA circumference 65.87 (CI: 65.02, 66.71) cm and hip cir- were high and significant at r > 0.7 and P < 0.0001 cumference 82.48 (CI: 80.97, 83.99) cm participated. (Figs. 3a and 4a). The Bland-Altman plots for FFM, FM, Cronbach’s alpha values for the two measurements of and %BF showed a random nature of spread with no de- saliva D O enrichment were high (0.999, 0.997 and tectable significant negative bias for FFM, FM and % BF 0.996 for children, young and older adolescents, respect- values estimated by DDM and BIA among the different ively) (Table 2). age groups (Figs. 2b, 3b and 4b). The correlation coefficients for deuterium enrichment DDM and BIA exhibited generally narrower limits of at 3 and 4 h were high and positive among children, agreement for FFM, FM or % BF among children and young and older adolescents at r = 0.998, 0.995, and young adolescents than among older adolescents (Fig. 2b, 0.993 respectively (Fig. 1a). The Bland-Altman plots 3b, and 4b). Older adolescents (15–19-years) exhibited showed random nature of spread with no detectable pro- the largest mean differences for FFM (− 2.84 kg), FM portional bias for saliva D O enrichment at 3 and 4 h (2.84 kg), and %BF (5.01) while young adolescents (10– Table 1 Participants’ characteristics Mean (95% Confidence Interval) Characteristic Children (8–9 years) Young adolescents (10– Older adolescents (15– Overall 14 years) 19 years) N 16 112 67 195 Male 7 51 26 84 Female 9 61 41 111 Age 8.34 (8.11 to 8.64) 11.80 (11.57 to 12.04) 17.39 (17.09 to 17.69) 13.44 (12.98 to 13.90) Weight (kg) 28.31 (25.88 to 30.75) 40.40 (38.74 to 42.05) 55.55 (53.48 to 57.62) 44.61 (42.92 to 46.31) Height (m) 1.31 (1.28 to 1.35) 1.48 (1.47 to 1.50) 1.62 (1.60 to 1.64) 1.51 (1.50 to 1.53) BMI (kg/m ) 16.30 (15.65 to 16.95) 18.16 (17.69 to 18.63) 21.23 (20.53 to 21.93) 19.07 (18.64 to 19.50) Waist circumference (cm) 58.56 (56.83 to 60.28) 65.07 (64.06 to 66.08) 68.94 (67.62 to 70.26) 65.87 (65.02 to 66.71) Hip circumference (cm) 68.43 (66.25 to 70 .60) 79.20 (77.70 to 80.71) 91.32 (89.17 to 93.47) 82.48 (80.97 to 84.26) Waist height ratio 0.44 (0.44 to 0.46) 0.44 (0.43 to 0.44) 0.43 (0.42 to 0.44) 0.44 (0.43 to 0.44) Waist hip ratio 0.86 (0.84 to 0.88) 0.82 (0.82 to 0.83) 0.76 (0.74 to 0.78) 0.80 (0.80 to 0.81) 3 h deuterium enrichment 722.86 (626.83 to 796.77 (772.52 to 821.03) 790.85 (744.72 to 836.99) 788.67 (766.49 to (ppm) 818.88) 810.86) 4 h deuterium enrichment 720.06 (622.18 to 799.78 (775.74 to 823.83) 798.69 (753.26 to 844.11) 792.87 (770.84 to (ppm) 817.94) 814.89) DDM Total body water (litres) 17.18 (15.90 to 18.47) 24.97 (24.07 to 25. 87) 32.05 (30.75 to 33.35) 26.76 (25.84 to 27.68) DDM Total body water (%) 60.92 (59.38 to 62.47) 62.29 (61.44 to 63.14) 58.25 (56.21 to 60.28) 60.79 (59.90 to 61.68) DDM Fat free mass (kg) 23.48 (21.72 to 25.23) 34.11 (32.88 to 35.34) 43.78 (42.00 to 45.56) 36.56 (35.30 to 37.82) BIA Fat free mass (kg) 24.13 (22.10 to 26.16) 33.89 (32.72 to 35.06) 46.62 (45.05 to 48.19) 37.46 (36.13 to 38.79) DDM Fat mass (kg) 4.84 (3.90 to 5.77) 6.29 (5.61 to 6.96) 11.77 (9.97 to 13.58) 8.05 (7.23 to 8.87) BIA Fat mass (kg) 4.18 (3.48 to 4.88) 6.51 (5.80 to 7.22) 8.93 (7.54 to 10.32) 7.15 (6.50 to 7.81) DDM Fat (%) 16.77 (15.65 to 16.95) 14. 90 (13.74 to 16.06) 20.43 (17.64 to 23.71) 16.95 (15.74 to 18.17) BIA Fat (%) 14.61 (12.83 to 16.39) 15.34 (14.21 to 16.45) 15.42 (13.32 to 17.52) 15.30 (14.34 to 16.27) Impedance 660.00 (626.16 to 596.63 (581.35 to 611.90) 515.04 (500.91 to 529.16) 573.79 (561.67 to 693.84) 585.92) Ndagire et al. BMC Public Health (2018) 18:687 Page 5 of 9 Table 2 Cronbach’s alpha values of the two readings for the enrichment measurements were reproducible among the different age groups study population. The high Cronbach’s alpha value (> Age category Cronbach’s Alpha 0.9) among all studied age groups indicates very good re- peatability of the FTIR spectroscopy saliva D O enrich- 8–9 years 0.999 ment measurements among children, young and older 10–14 years 0.997 adolescents in Uganda. The FTIR spectroscopy instru- 15–19 years 0.996 ment can, therefore, provide reliable measures for D O saliva enrichment and thus suitable for validation of 14-years) exhibited lowest mean differences for FFM other body composition assessment techniques for more (0.22 kg), FM (− 0.22 kg) and %BF (− 0.44) (Table 3). accurate assessment of body composition among chil- Furthermore, the mean differences between DDM and dren and adolescents in Uganda. Furthermore, the FTIR BIA for measures of FFM, FM and %BF for older spectroscopy technique has several advantages in asses- adolescents exhibited largest 95% confidence intervals sing body composition including simplicity to carry out, compared to those for children and young adolescents. minimal subject cooperation requirements, acceptability The Bland Altman plots for FFM, FM, and %BF for older in all age groups [28], non-invasiveness, relatively low adolescents exhibited largest limits of agreement cost, easy administration of tracers, and easy collection compared those of children and young adolescents of samples [29]. (Fig. 2b, 3b and 4b). In this study, the ability of the inbuilt equations from the Tanita SC-331S BIA instrument to assess body com- Discussion position of children and adolescents in Uganda by using Prior to this work, no published studies were found on DDM as a reference method also was investigated. Prior the reliability of FTIR saliva D O enrichment measure- to this work, no published studies were found compar- ments among populations in Uganda. In this study, the ing the body composition estimates obtained by BIA to reliability of FTIR spectroscopy measurements among those obtained by DDM among children and adolescents children and adolescents in Uganda was assessed. The in Uganda. Since estimates for body composition had high and positive correlation between 3 and 4-h FTIR varying agreement across the studied age groups, DDM spectroscopy measurements is indicative of similarity and BIA are generally not interchangeable across chil- and reproducibility of the two sets of measurements. dren and adolescents in Uganda. The none-statistically The Bland-Altman plots that showed no apparent trend significantly different (P > 0.05) FFM, FM and %BF mea- in error differences between the measurements taken sures by DDM and BIA among children and young ado- after 3 and those taken after 4 h imply that saliva D O lescents imply possibility for agreement between the two Fig. 1 Regression and Bland-Altman plots for saliva D O enrichment among children (left), young adolescents (middle) and older adolescents (right) Ndagire et al. BMC Public Health (2018) 18:687 Page 6 of 9 Table 3 Body composition mean values (CI), mean difference (CI) and P-values between DDM and BIA among different age groups Body composition means DDM BIA Mean difference P-value Children (8–9 years) FFM (kg) 23.48 (21.72–25.23) 24.13 (22.10–26.16) −0.657 (−1.474–0.160) 0.1071 FM (kg) 4.84 (3.90–5.77) 4.18 (3.48–4.88) 0.657 (− 0.160–1.474) 0.1071 %BF 16.77 (14.66–18.88) 14.61 (12.83–16.39) 2.157 (−0.397–4.711) 0.0920 Young adolescents (10–14 years) FFM (kg) 34.11 (32.88–35.34) 33.89 (32.72–35.06) 0.224 (− 0.111–0.559) 0.1876 FM (kg) 6.29 (5.61–6.96) 6.51 (5.80–7.22) −0.224 (− 0.559–0.111) 0.1876 %BF 14. 90 (13.74–16.06) 15.34 (14.21–16.45) −0.436 (− 1.239–0.367) 0.2846 Older adolescents (15–19 years) FFM (kg) 43.78 (42.00–45.56) 46.62 (45.05–48.19) −2.841 (− 3.983 - -1.699) < 0.0001 FM (kg) 11.77 (9.97–13.58) 8.93 (7.54–10.32) 2.841 (1.699–3.983) < 0.0001 %BF 20.43 (17.64–23.71) 15.42 (13.32–17.52) 5.006 (3.068–6.944) < 0.0001 methods in these age categories. For children and young validate 2 portable BIA devices; the Inbody 230 and the adolescents, the generally narrow limits of agreement, Tanita BC-418 for body composition assessment in the small mean discrepancies (biases) for the FFM, FM healthy Taiwanese school-age children, Bland-Altman and %BF estimates and their narrow 95% confidence in- analysis showed clinically acceptable agreement between tervals of means imply that DDM and BIA estimates for the Inbody 230 device and DEXA for FFM measure- FFM, FM, and %BF agree and can be used interchange- ments [15]. ably for either FM, FFM, or %BF for these age categories On the other hand, the statistically significantly differ- in Uganda. These findings are similar to those by Mehta ent mean values (P < 0.05) for FFM, FM and %BF among and others who found agreement between BIA and older adolescents imply no possibility for agreement be- DDM for FFM, FM and %BF among children 14 years of tween the two methods. The wide limits of agreement age or younger with Intestinal Failure [23]. In a study to for FFM, FM, or %BF exhibited by Bland Altman plots Fig. 2 Regression and Bland-Altman plots for FFM (left), FM (middle) and % body fat (right) determined by DDM and BIA among children Ndagire et al. BMC Public Health (2018) 18:687 Page 7 of 9 Fig. 3 Regression and Bland-Altman plots for FFM (left), FM (middle) and % body fat (right) determined by DDM and BIA among young adolescents for older adolescents, the big mean discrepancies interchangeable for either FM, FFM, or %BF among (biases) for the FFM, FM and %BF estimates and their older adolescents (15–19 years) in Uganda. Similar to wide 95% confidence intervals in this age group imply this study’s findings where BIA overestimated FFM limited agreement between the two methods. This re- among older adolescent and underestimated their FM veals that DDM and BIA are not directly and %BF are those by Resende and others who reported Fig. 4 Regression and Bland-Altman plots for FFM (left), FM (middle) and % body fat (right) determined by DDM and BIA among old adolescents Ndagire et al. BMC Public Health (2018) 18:687 Page 8 of 9 that BIA overestimated the measures of FFM and under- composition in resource-poor countries that cannot af- estimated the measures of FM compared to those pro- ford four-compartment (gold standard) techniques. vided by DDM among obese adolescents in Brazil [5]. The other results of the study showed that DDM and Resende and others reported a high, positive and signifi- BIA can be used interchangeably for FFM, FM, and %BF cant correlation between FFM and FM values deter- for children and young adolescents aged 8–14 years in mined by DDM and BIA but there was no agreement Uganda but not interchangeable for the assessment of between the two methods among obese adolescents [5] body composition in older adolescents aged 15–19 years as was the case for older adolescents in this study. In a in Uganda. For that reason, among older adolescents in study to validate predictive equations of BIA to FFM es- Uganda, BIA is not a valid measure for body compos- timation in army cadets aged 17–24 years, Langer and ition, so deriving population-specific BIA equations may others observed significant differences between FFM be a suitable approach for assessing body composition. values from 8 predictive BIA equations and no good The study, therefore, revealed BIA’s limitations in asses- agreement with DXA [11] Also, among healthy Indian sing body composition among children and adolescents children and adolescents aged 5–18 years, there was no in Uganda. agreement between BIA and DXA in assessment of body Abbreviations composition [13]. A possible explanation for the discrep- 2 %BF: Body fat percentage; H: Deuterium; ADP: Air displacement ancy of body composition among older adolescents by plethysmography; BIA: Bioelectrical impedance analysis; BMI : Body mass index; D O: Deuterium oxide; DDM: Deuterium dilution method; DEXA: Dual- the BIA system’s inbuilt prediction equations is that they energy x-ray absorptiometry; FFM: Fat-free mass; FM: Fat mass; FTIR: Fourier are normally based on Western European or North transform infrared; IAEA: International atomic energy agency; LMICs: Low and American populations, which may differ in body com- middleiIncome countries; NCDs : Non-communicable diseases; TBW: Total body water; UWW : Underwater weighing; VD: Dilution space; WHO: World position and proportion when compared to the popula- Health Organization tion under study [30]. Growth involves the deposition of both fat mass (FM) and fat-free mass (FFM) components Acknowledgements The authors are grateful to the children and adolescents who participated in and human body composition is ethnicity dependent this study, their parents/guardians, schools’ administration and teachers and [31]. No literature was found regarding the age- and the entire research team. sex-related pattern of changes in body composition for Funding populations in Uganda. The International Atomic Energy Agency (IAEA) offered the study support While the study was the first of its kind among under project number UGA6017. populations in Uganda, it was not without limitations. DDM was used as the reference method which, Availability of data and materials The datasets used and/or analyzed during the current study are available although widely validated as a reliable estimate, is not from the corresponding author on reasonable request. a gold standard for body composition. Ideally, a four-component model would have been used as the ref- Authors’ contributions CTN, DI, JHM, JEA, and DN conceived, designed, and revised the manuscript. erence method, but this was not possible in our study CTN did the literature search. SMAS, RB, and BO did the statistical analysis. All setting. While DDM has the advantage that it is rela- authors read and approved the final manuscript. tively easy to perform, it is not without limitations: one Ethics approval and consent to participate assumption is the hydration of FFM, which may vary The purpose and objectives of the study were carefully explained to each among persons by age, sex, maturation and ethnicity participant and their parents. Informed consent to the study was obtained and to estimate FFM from TBW, age, and sex-specific from participants’ parent/guardian to affirm their willingness or not. The parents or guardians of participants provided consent to allow their children hydration fractions were used [6]. But the hydration of to take part in the study while participants signed assent accepting to FFM values used for computation of TBW to estimate participate in the study. Ethical clearance to engage human subjects was FFM, FM, and %BF were not Uganda specific. Higher obtained from Makerere University School of Biomedical Sciences Higher Degrees, Research and Ethics Committee and Uganda National Council for hydration factors have been observed among African Science and Technology under reference numbers: SBS 291 and HS 1950 American adults compared to whites using a respectively. four-component model [32]. However, there is no infor- Competing interests mation on the hydration factors of FFM for Ugandan The authors declare that they have no competing interests. populations. Publisher’sNote Conclusions Springer Nature remains neutral with regard to jurisdictional claims in The reliability of the FTIR spectroscopy saliva D O en- published maps and institutional affiliations. richment measurements was very good among the stud- Author details ied population. This technique can be used as a School of Food Technology, Nutrition and Bio-engineering, Makerere reference technique in the validation of field techniques 2 University, Kampala, Uganda. Department of Biochemistry and Sports like BIA for more accurate estimation of body Science, Makerere University, Kampala, Uganda. Department of Statistical Ndagire et al. BMC Public Health (2018) 18:687 Page 9 of 9 Methods and Actuarial Science, School of Statistics, Makerere University, stud 2015. https://www-users.york.ac.uk/~mb55/meas/sizemeth.htm. 4 5 Kampala, Uganda. Centre MURAZ, Bobo-Dioulasso, Burkina Faso. Division of Accessed 7 Feb 2018. Nutritional Sciences, University of Illinois, Urbana-Champaign, USA. 19. Willet W. Nutritional epidemiology. Oxford: Oxford University Press; 1998. Department of Food Science and Human Nutrition, University of Illinois, 20. Hobart JC, Cano SJ, Warner TT, Thompson AJ. What sample sizes for Urbana-Champaign, USA. reliability and validity studies in neurology? J Neurol. 2012;259:2681–94. https://doi.org/10.1007/s00415-012-6570-y. Received: 3 November 2017 Accepted: 29 May 2018 21. Bosy-Westphal A, Later W, Hitze B, Sato T, Kossel E, Glüer C-C, et al. Accuracy of bioelectrical impedance consumer devices for measurement of body composition in comparison to whole body magnetic resonance imaging and dual X-ray absorptiometry. Obes Facts. 2008;1:319–24. https://doi.org/ References 10.1159/000176061. 1. Celermajer DS, Chow CK, Marijon E, Anstey NM, Woo KS. Cardiovascular 22. Li YC, Li CI, Lin WY, Liu CS, Hsu HS, Lee CC, et al. Percentage of body fat disease in the developing world. J Am Coll Cardiol. 2012;60:1207–16. assessment using bioelectrical impedance analysis and dual-energy X-ray https://doi.org/10.1016/j.jacc.2012.03.074. absorptiometry in a weight loss program for obese or overweight Chinese 2. Raj M. Obesity and cardiovascular risk in children and adolescents. Indian J adults. PLoS One. 2013;8:2–8. https://doi.org/10.1371/journal.pone.0058272. Endocrinol Metab. 2012;16:13. https://doi.org/10.4103/2230-8210.91176. 23. Mehta NM, Raphael B, Guteirrez IM, Quinn N, Mitchell PD, Litman HJ, et al. 3. WHO. Obesity and overweight @ WwwWhoInt 2017. http://www.who.int/ Comparison of body composition assessment methods in pediatric es/news-room/fact-sheets/detail/obesity-and-overweight. Accessed 1 June intestinal failure. J Pediatr Gastroenterol Nutr. 2014;59:99–105. https://doi. org/10.1097/MPG.0000000000000364. 4. Akindele MO, Phillips JS, Igumbor EU. The relationship between body fat 24. Shafer KJ, Siders WA, Johnson LK, Lukaski HC. Validity of segmental multiple- percentage and body mass index in overweight and obese individuals in an frequency bioelectrical impedance analysis to estimate body composition of urban african setting. J Public Health Africa. 2016;7. https://doi.org/10.4081/ adults across a range of body mass indexes. Nutrition. 2009;25:25–32. jphia.2016.515. https://doi.org/10.1016/j.nut.2008.07.004. 5. Resende CMM, Camelo Júnior JS, Vieira MNCM, Ferriolli E, Pfrimer K, 25. Widen EM, Strain G, King WC, Yu W, Lin S, Goodpaster B, et al. Validity of Perdoná GSC, et al. Body composition measures of obese adolescents by bioelectrical impedance analysis for measuring changes in body water and the deuterium oxide dilution method and by bioelectrical impedance. percent fat after bariatric surgery. Obes Surg. 2014;24:847–54. https://doi. Brazilian J Med Biol Res. 2011;44:1164–70. https://doi.org/10.1590/S0100- org/10.1007/s11695-014-1182-5. 879X2011007500122. 26. Ramírez E, Valencia ME, Bourges H, Espinosa T, Moya-Camarena SY, Salazar 6. Srichan W, Yamborisut U, Kijboonchoo K, Thasanasuwan WA. G, et al. Body composition prediction equations based on deuterium oxide Comparison of bioelectrical impedance analysis with deuterium dilution dilution method in Mexican children: a national study. Eur J Clin Nutr. 2012; technique for body fat assessment in school-age children. J Public 66:1099–103. https://doi.org/10.1038/ejcn.2012.89. Health (Bangkok). 2014;44:223–36. 27. Chirita-Emandi A, Dobrescu A, Papa M, Puiu M, Emandi AC. Reliability of 7. Molina KM, Molina KM, Goltz HH, Kowalkouski MA, Hart SL, Latini D, et al. measuring Subcutaneous fat Tissue thickness using Ultrasound in non- Reliability and validity, Encycl. Behav. Med. New York, NY: Springer New athletic young adults MEASURING SUBCUTANEOUS FAT TISSUE USING York; 2013. p. 1643–4. https://doi.org/10.1007/978-1-4419-1005-9_1549. ULTRASOUND. Maedica -a J Clin med Maedica a J Clin med J Clin med 8. Meissl K, Smidt E, Tintner J. Reproducibility of fourier transform infrared Maedica a. J Clin Med. 2015;10:204–9. spectra of compost, municipal solid waste, and landfill material. Appl 28. IAEA. Nutritional and Health-Related Environmental Studies (NAHRES). Using Spectrosc. 2008;62:190–6. https://doi.org/10.1366/000370208783575537. nuclear techniques to assess the role of nutrition-sensitive agri-food systems 9. Bland MJ, Altman DG. Statistical methods for assessing agreement between in improving diet, health and nutritional status of vulnerable populations. two methods of clinical measurement. Lancet London J Onwhyn, 1823- 2015. http://www-naweb.iaea.org/nahu/NAHRES/crp/e43029.html. Accessed Oxford Elsevier 1986;327:307–10. doi:https://doi.org/10.1016/S0140- 14 Feb 2018. 6736(86)90837-8. 29. Lee SY, Gallagher D. Assessment methods in human body composition. 10. Böhm A, Heitmann BL. The use of bioelectrical impedance analysis for body Curr Opin Clin Nutr Metab Care. 2008;11:566–72. https://doi.org/10.1097/ composition in epidemiological studies. Eur J Clin Nutr. 2013;67:S79–85. MCO.0b013e32830b5f23.Assessment. https://doi.org/10.1038/ejcn.2012.168. 30. Wickramasinghe V, Lamabadusuriya S, Cleghorn G, Davies P. Assesment of 11. Langer RD, Borges JH, Pascoa MA, Cirolini VX, Guerra-Júnior G, Gonçalves body composition in Sri Lankachildren: validation of a biolectrical EM. Validity of bioelectrical impedance analysis to estimation fat-free mass impedance prediction equation. Eur J Clin Nutr. 2008;62:1170–7. https://doi. in the army cadets. Nutrients. 2016;8 https://doi.org/10.3390/nu8030121. org/10.1016/0377-8401(90)90166-6. 12. Dehghan M, Merchant AT. Is bioelectrical impedance accurate for use in 31. Xiong K-Y, He H, Zhang Y-M, Ni G-X. Analyses of body composition large epidemiological studies? Nutr J. 2008;7:26. https://doi.org/10.1186/ charts among younger and older Chinese children and adolescents 1475-2891-7-26. aged 5 to 18 years. BMC Public Health. 2012;12:835. https://doi.org/10. 13. Chiplonkar S, Kajale N, Ekbote V, Mandlik R, Parthasarathy L, Khadilkar V, et 1186/1471-2458-12-835. al. Validation of bioelectric impedance analysis against dual-energy X-ray 32. Prins M, Hawkesworth S, Wright A, Fulford a JC, LM a J, Prentice a M, et al. absorptiometry for assessment of body composition in Indian children aged Use of bioelectrical impedance analysis to assess body composition in rural 5 to 18 years. Indian Pediatr. 2017;54:919–24. Gambian children. Eur J Clin Nutr. 2008;62:1065–74. https://doi.org/10.1038/ 14. Haroun D, Croker H, Viner RM, Williams JE, Darch TS, Fewtrell MS, et al. sj.ejcn.1602830. Validation of BIA in obese children and adolescents and re-evaluation in a longitudinal study. Obesity. 2009;17:2245–50. https://doi.org/10. 1038/oby.2009.98. 15. Lee L-W, Liao Y-S, Lu H-K, Hsiao P-L, Chen Y-Y, Chi C-C, et al. Validation of two portable bioelectrical impedance analyses for the assessment of body composition in school age children. PLoS One. 2017;12:e0171568. https:// doi.org/10.1371/journal.pone.0171568. 16. Wang L, Hui SS. Validity of four commercial bioelectrical impedance scales in measuring body fat among Chinese children and adolescents. Biomed Res Int. 2015;2015:1–8. https://doi.org/10.1155/2015/614858. 17. Anthoine E, Moret L, Regnault A, Sébille V, Hardouin J-B. Sample size used to validate a scale: a review of publications on newly-developed patient reported outcomes measures. Health Qual Life Outcomes. 2014;12:2. https:// doi.org/10.1186/s12955-014-0176-2. 18. Bland M. How can I decide the sample size for a study of agreement between two methods of measurement? Freq asked quest des anal Meas

Journal

BMC Public HealthSpringer Journals

Published: Jun 4, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off