AMERICAN JOURNAL OF HUMAN BIOLOGY 27:349–357 (2015)

Original Research Article

Development, Validation, and Use of a Semi-Quantitative Food Frequency Questionnaire for Assessing Protein Intake in Papua New Guinean Highlanders AYAKO MORITA,1,2* KAZUMI NATSUHARA,3 ERIKO TOMITSUKA,1,4 SHINGO ODANI,5 JUN BABA,6 KIYOSHI TADOKORO,1,7 KATSURA IGAI,1,8 ANDREW R. GREENHILL,9,10 PAUL F. HORWOOD,9 KEVIN W. SOLI,9 SUPARAT PHUANUKOONNON,9,11 PETER M. SIBA,9 AND MASAHIRO UMEZAKI1 1 Department of Human Ecology, The University of Tokyo, Tokyo 113-0033, Japan 2 Department of Health Promotion, Tokyo Medical and Dental University, Tokyo 113-0034, Japan 3 Faculty of Nursing, The Japanese Red Cross Akita College of Nursing, Akita 010-1493, Japan 4 Niigata University of Pharmacy and Applied Life Sciences, Niigata 956-8603, Japan 5 Faculty of Letter, Chiba University, Chiba 263-8522, Japan 6 Tokyo Metropolitan University, Tokyo 192-0397, Japan 7 Faculty of International Resource Sciences, Akita University, Akita 010-8502, Japan 8 Department of International Health, Nagasaki University, Nagasaki 852-8523, Japan 9 Papua New Guinea Institute of Medical Research, Goroka, EHP 441, Papua New Guinea 10 School of Applied and Biomedical Sciences, Federation University, Churchill Campus, Victoria 3841, Australia 11 Infection and Immunity Division, The Walter and Eliza Hall Institute of Medical Research, Melbourne, Victoria 3052, Australia

Objectives: The aim of this article was to develop a semi-quantitative food frequency questionnaire (FFQ) and evaluate its validity to estimate habitual protein intake, and investigate current dietary protein intakes of Papua New Guinea (PNG) Highlanders. Methods: A 32-item FFQ was developed and tested among 135 healthy male and female volunteers. The FFQestimated daily total and animal protein intakes were compared with biomarkers and 3-day Weighed Food Records (WFR) by correlation analyses, Bland–Altman plot analyses and joint classification analyses. Results: The FFQ-estimated total protein intake significantly correlated with urinary nitrogen in the first morning void after adjusting urinary creatinine concentration (r 5 0.28, P < 0.01) and the FFQ-estimated animal protein intake significantly correlated with the hair d15N (Spearman’s r 5 0.34, P < 0.001). The limits of agreement were 62.39 Z-score residuals for total protein intake and 62.19 Z-score for animal protein intake, and intra-individual differences increased as protein intake increased. The classification into the same and adjacent quartiles was 66.0% for total protein intake and 73.6% for animal protein intake. Median daily total and animal protein intake estimates from the FFQ and the 3-day WFR showed a good agreement with differences of 0.2 and 4.9 g, respectively. None of the studied communities in the PNG Highlands met the biologically required protein intake; although the community closer to an urban center showed higher protein intake than the more remote communities. Conclusions: The newly developed 32-item FFQ for PNG Highlanders is applicable for evaluation of protein intake C 2014 Wiley Periodicals, Inc. V at the individual level. Am. J. Hum. Biol. 27:349–357, 2015. People living in the highlands of Papua New Guinea (PNG) have a heavy dietary dependence on sweet potatoes, which have a low protein content (Bourke and Allen, 2010; Muntwiler and Shelton, 2001). Although protein is an essential nutrient for muscle synthesis and maintenance throughout life, many studies conducted in the PNG highlands has reported that daily protein intake in the adult population is lower than the WHO-recommended dietary intake (Bailey and Whiteman, 1963; Harvey and Heywood, 1983; Koishi, 1990; Ulijaszek et al., 1987). However, this population has consistently displayed apparent protein adequacy and good muscular development. The average serum total protein level is normal and comparable to Japanese (Koishi, 1990); average body mass index (BMI) is within the optimal range proposed for Pacific Islanders (18.5–27 kg/m2, Rush et al., 2009) (Bailey and Whiteman, 1963; Koishi, 1990); average mid-upper arm circumference (MUAC) is above the cut-off points for adults in developing countries (22 cm in women, 23 cm in men, Tang et al., 2013;Yamauchi, 2001); and superior scores over Australians in bicycle ergometry and Harvard Pack test have been reported (Sinnett and Solomon, 1968). Recycling of urea–nitrogen in the intestine, a reduced protein turnover, and supplementation of nitrogen fixed by intestinal bacteria have been proposed in PNG Highlanders, which may contribute to this apparent discrepC 2014 Wiley Periodicals, Inc. V

ancy between the low protein diet and nutritional outcome (Bergersen and Hipsley, 1970; Koishi, 1990; Luyken et al., 1964). However, to date, there is no definitive explanation. Despite recent technological developments, there is still no reliable method of estimating the habitual protein intake of PNG Highlanders at an individual level. Earlier nutrition studies in PNG weighed all the food consumed over a 3–8 day period (Bailey and Whiteman, 1963; Harvey and Heywood, 1983; Koishi, 1990; Norgan et al., 1974; Sinnett, 1972; Ulijaszek et al., 1987; Umezaki et al., 1999) or asked the subjects to recall all food consumed in the

Contract grant sponsor: Cabinet Office, Government of Japan (Funding Program for Next Generation World-Leading Researchers); Contract grant number: LS024. Conflicts of Interest: The authors declare no conflict of interest. This study was approved by the Institutional Review Board at Papua New Guinea Institute of Medical Research (1025), the National Research Institute of Papua New Guinea, the Medical Research Advisory Committee of Papua New Guinea (11.16), and the Research Ethics Committee at the Graduate School of Medicine, the University of Tokyo (3391). *Correspondence to: Ayako Morita, Department of Human Ecology, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. E-mail: [email protected] Received 18 May 2014; Revision received 18 August 2014; Accepted 6 October 2014 DOI: 10.1002/ajhb.22647 Published online 3 November 2014 in Wiley Online Library (wileyonlinelibrary.com).

350

A. MORITA ET AL.

last 24 h for 2 days (Gibson, 1998). This may reflect intake on specific days but may not be representative of usual intake over a long period. For example, protein intake is known to be temporally highly variable in many Highlanders, where little protein may be consumed for long periods then a relatively high protein intake may occur through consumption of pigs at a customary ceremony (Koishi, 1990). If conducted over weeks, a weighed food record (WFR) and a 24-h dietary recall should provide a reliable protein intake level of individuals. However, that is burdensome on both respondents and researchers, and thus not feasible, especially for large-scale studies. A food frequency questionnaire (FFQ) is a relatively simple and inexpensive dietary method for measuring an individual’s habitual dietary intake (Willett, 1986). It consists of a list of foods and options to indicate the usual frequency of consumption of each food on the list. If options to indicate usual portion size are added to a FFQ, it is possible to collect semi-quantitative dietary data. These characteristics make FFQs suitable for identifying individuals on a low-protein diet in the PNG highlands. However, there is no validated FFQ that is culturally sensitive to the dietary habits of PNG Highlanders. It is important to use a FFQ specifically tailored to a target population, because what and how food is consumed can vary greatly according to food environment and social organization (Willet, 1986). It is also important to select food items that are most informative for the study objective, because a lengthy FFQ can cause fatigue and boredom in the subjects (Willet, 1986). Finally, it is important to examine the performance of the FFQ in comparison with independent standards, such as nutrient biomarkers and WFR, to examine potential errors associated with self-reporting. This article is the first report of the development of a culturally appropriate FFQ designed specifically to estimate individual habitual protein intake in PNG Highlanders. We then investigated its validity against biomarkers of protein intake and against the results of a 3-day WFR (3d-WFR). Finally, we evaluated the level of protein intake in three communities in PNG Highlands with different distances from urban centers. MATERIAL AND METHODS Development of the FFQ for PNG Highlanders A list of major protein foods in PNG Highlanders’ diet was previously constructed from a 7d-WFR of 25 male and female Highlanders aged 10 years and above residing in two villages in Asaro Valley, in the Eastern Highlands Province. Details of the study sites and participants have been described elsewhere (Natsuhara, 2002). Foods were included in the questionnaire if they made a more than a 1% contribution to total protein intake or ranked in the top 20 according to the average intake per occasion. In total, we selected 31 food items that covered 98.2% of the total protein intake among the participants, which we ensured were representative of foods currently eaten by PNG Highlanders by observing dietary habits in the communities around Goroka (Eastern Highlands Province) and Tari (Hela Province) in August 2012. For each food item on the list, we prepared response categories for participants to indicate usual consumption frequency per day or week, or the total number of times a food was consumed in the past month. For snack bar foods, we prepared an open-ended response question for American Journal of Human Biology

participants to list all the items they had consumed and the total number consumed per item in the past month. Usual portion size was specified using realistic food samples for root crops, animals, and leafy vegetables, body parts for fish (the use of the hand, arm, and leg are common ways to describe the size of fish in PNG Highland society) and natural units (e.g. piece, plate, small/ medium/large) for the rest of the food items. Interpretation of natural units can vary from site to site, so we gathered food samples from markets and gardens in each site, and used the average as the usual portion size. For commonly shared foods, such as rice, noodles, and canned meat/fish, each participant was asked to describe the usual packet or can size and usual number of people that shared the food per occasion. Because males require more energy than females due to their larger body sizes, we examined whether they differed in serving portions for sharing types of food based on a 3d-WFR collected in Frigano in February 2012 (a more detailed description is provided in the measurement section below). We observed that adult males received a 1.4-fold greater portion of rice, 1.95-fold greater portion of canned food, 1.5-fold greater portion of noodles, and 1.4-fold greater portion of meat than adult females from the same household, on average. We calculated the median portion ratio, and used it (male:female 5 1.5:1) to adjust the typical serving size of rice, canned food, noodles and meat. Habitual total protein intake in the preceding month was calculated as the sum of the products of consumption frequency, portion size, and protein content for each food item in the preceding month divided by 30 days. We used published data for the protein contents of specific foods in this study. For sweet potato, we used PNG-specific data (Umezaki et al., 2001). Where possible, the Pacific Islands food composition tables (Dignan et al., 2004) were used for other foods; if not listed there, we referred to the Australian Food, Supplement & Nutrient Database (Food Standard Australia New Zealand, 2008) or Standards Tables of Food Composition from Japan (Japanese Ministry of Education, Culture, Sports, Science and Technology, 2005). Protein intake from different food categories was also calculated. The food items were classified into animal-source or vegetable-source. Animal-source foods included meat, poultry, and fish, while vegetable-source foods included vegetables, fruits, and nuts. Mixed food was also classified as animal- or vegetable-source, depending on the main ingredient. For example, the main ingredient of lamb stew was lamb (meat); thus, we classified it as an animal-source food. The food items were also classified into subsistence food or store-bought food. Subsistence food was defined as food generally produced in the Highland villages: pork, game meat, freshwater fish, root crops, corn, green leaves, other vegetables, nuts, and sweet banana. Store-bought food was defined as food that required ingredients made and purchased at retail stores in general: chicken, lamb/mutton, canned food, saltwater fish, egg, rice, bakery, noodles, and all food from the snack bar. Usual protein intake from each food category in the preceding month was calculated in the same way as total protein intake. Study sites and participants This study was conducted in three Highland communities: Frigano (Eastern Highlands Province), Wenani (Tari Basin, Hela Province), and Levani (Levani Valley, Hela

PROTEIN INTAKE IN PAPUA NEW GUINEAN HIGHLANDERS

Province). The residents were mostly subsistence farmers who generate only small amounts of money by selling garden crops or pigs in the markets to purchase store food (Umezaki et al., 1999; Natsuhara and Ohtsuka, 1999) but the degrees of exposure to market economy differed among the communities. The residents in Frigano, situated at an altitude of 1,600 m, have the greatest economic and geographic access to an urban center and market economy. Public transport is available to Goroka town, the provincial capital, where a system to connect smallscale farmers to buyers of coffee, livestock, and vegetables has been established and many commercial facilities (e.g. factories, offices, supermarkets) exist. Wenani, situated at an altitude of 1,650 m, is more distant from its nearest urban center than is Frigano. It is a 30-min walk to the nearest public transport stop and another 30 min travel by public transport to Tari town. The town is a provincial capital, but there are few commercial facilities such as exist in Goroka. Levani, at an altitude of 2,300 m, is situated in an even more remote area. It is an 8-h walk and requires crossing a steep, narrow mountain path, in addition to driving the Highlands Highway, the country’s major land route, for 2 h, to reach Tari town. Each community leader called the residents to gather in an open space to hear about the study. Inclusion criteria for the FFQ validation study were aged 10 years and over, with no severe illness. Between February and March in 2012–2013, in total, 136 eligible persons offered to participate voluntarily and provided written informed consent. Measures Food frequency questionnaire. We conducted the FFQ survey by means of face-to-face interviews about consumption frequency and typical portion sizes in the preceding month. We observed pandanus nuts (Pandanus julianeti) to be commonly eaten in Wenani and Levani at the time of the survey and identified it as an important source of protein in these communities. Thus, we added a question regarding pandanus nuts consumption in the two communities, resulting in a 32-item FFQ. For weighing food samples, we used digital scales (Dretec, KS-208, precision of 1 g for 0–2.0 kg) and spring scales (Sanko, 1 kg, 5 kg).

Urinary nitrogen as a total protein intake biomarker. Protein contains 16% nitrogen, which is eliminated primarily via urine as urea (Bingham and Cummings, 1985), and urinary nitrogen (UN) excretion amounts reflect total dietary protein intake (Bingham, 2003). Given logistical barriers in the field, we decided to collect the first morning void. Single overnight UN concentration has been shown to correlate with the mean of 24-h UN concentrations over 8 days with correlation coefficients of 0.29 in an urban Caucasian population (Fuller et al., 2005). We visited each household to provide participants with a 200-ml paper cup labelled with an identifier and instructed them to collect the first morning urine on the following day. The urine samples were brought to the researchers immediately after the participant had collected the sample, and transferred to a 4-mL tube with preservative (methyl-4-hydroxybenzoic acid). They were kept at room temperature until stored in the refrigerator

351

at 4 C in the nearest town. We brought them back to a laboratory at the University of Tokyo, where they were stored at –80 C until analyzed. UN concentration (mg/100 ml) was measured by the urease-indophenol method, and creatinine concentration (mg/100 ml) was determined by Jaffe’s method. Because protein metabolism is known to change in pregnancy (Fuller et al., 2005; Mojtahedi et al., 2001), we excluded urine samples of pregnant participants from the analyses. Hair d15N as an animal protein intake biomarker. Hair d15N (&) is a known biomarker of habitual animal protein intake in humans, where a higher value indicates greater animal protein intake (Bol and Pflieger, 2002; O’Connell and Hedges, 1999; Petzke et al., 2005, 2010; Reitsema, 2013; Yoshinaga et al., 1991, 1996). Each subject had a few strands of their head hair closest to the scalp cut with stainless-steel scissors, and placed in a resealable bag with an identifier. Afro-Caribbean hair grows approximately 1 cm in 1 month (Loussouarn, 2001). We trimmed the samples so that it included only 1.5–3 cm from the scalp, and analyzed them using an EA-IRMS system consisting of a Calro Erba NA1500 elemental analyzer, a Finnigan MAT ConFlo II interface, and a Finnigan MAT 252 or a delta V isotope ratio mass spectrometer. Stable isotopic compositions were reported with reference to atmospheric air [15N/14N]) and expressed in the d (delta) notation, where d15N(&) 5 [(R_sample/R_air) – 1] 3 1000. R is the absolute ratio of 15N to 14N isotopes. On the basis of the standard deviations of replicate analyses of the internal standard material, the uncertainty with each measurement was estimated at 0.4& for d15N. Weighed food record. To evaluate the FFQ-estimated quantity of an individual’s protein intake, direct weighing records of all food consumed were performed for three successive days in Frigano. On each day, we collected data on generic and brand names, food preparation methods, and weight of each food before and after the meal from 5 a.m. until the participants’ last meal. The researchers placed a chair near the participants’ houses (two to three houses per researcher at one time) and visited the house with a scale to weigh uncooked ingredients whenever smoke was coming out of the house or when the participants called, in addition to hourly visits. Seasonings were measured with tablespoons and teaspoons in the households. The food was weighed at the time of serving, and leftovers were weighed after the meal (e.g. bones) to calculate the actual consumed portion. For food consumption outside the house, we followed wherever possible. However, if a participant ate in our absence, we gathered the same food with a similar size from stores and/or households based on the participant’s report and weighed it. Protein content and intake were calculated in the same manner as for the FFQ. Demographics and anthropometry. We collected demographic information (gender, age, education) during faceto-face interviews, and anthropometric information (height, weight, BMI) according to the protocol of Weiner and Lourie (1981). Age was estimated from the demographic database constructed by Umezaki and Natsuhara American Journal of Human Biology

352

A. MORITA ET AL. TABLE 1. Characteristics of the study participants by sex and age-group WFR participants (n 5 26)a

FFQ participants (n 5 135) Male (n 5 73) Variable Heigh, cm (mean 6 SD)b Weight, kg (mean 6 SD)b BMI, kg/m2(mean 6 SD)b MUAC, cm (mean 6 SD)b Education (n, %)c None Primary (1–6 years) Secondary (7–12 years) Residence (n, %) Frigano Wenani Levani

Female (n 5 62)

Male (n 5 13)

Female (n 5 13)

Aged 201 (n 5 58)

Age 10–19 (n 5 15)

Aged 201 (n 5 49)

Age 10–19 (n 5 13)

Aged 201 (n 5 9)

Age 10–19 (n 5 4)

Aged 201 (n 5 11)

Age 10–19 (n 5 2)

161.4 6 5.6 62.3 6 7.9 23.9 6 2.5 27.2 6 2.7

– – – –

151.8 6 5.9 53.8 6 8.1 23.3 6 3.0 24.7 6 2.5

– – – –

163.1 6 5.7 66.5 6 11.8 24.9 6 3.9 28.4 6 3.8

– – – –

152.5 6 5.4 56.0 6 8.3 24.0 6 3.1 25.8 6 1.8

– – – –

47.4 36.8 15.8

6.7 86.7 6.6

57.1 24.5 18.4

30.8 46.2 23.0

25.0 50.0 25.0

0.0 100.0 0.0

18.2 54.5 27.3

50.0 0.0 50.0

22.4 51.7 25.9

26.7 46.7 26.7

32.7 40.8 26.5

23.1 38.5 38.5

100.0 0.0 0.0

100.0 0.0 0.0

100.0 0.0 0.0

100.0 0.0 0.0

FFQ: food frequency questionnaire; WFR: weighed food record; IQR: interquartile range; BMI: body mass index a In total, 31 FFQ participants completed WFR; however, 6 participated in an engagement feast where large chunks of meats (200 800 g per person, 500 g per person on average) were distributed. As a feast is not an every-day activity, feast participants were excluded from the analysis b Anthropometric characteristics of teenage groups (aged 10–19 years-old) are not presented because height, weight and MUAC are largely affected by age distribution and BMI is not an appropriate indicator to evalute degrees of obesity during the growth period. c Education information was missing for one adult male FFQ participant who also participated in the WFR; The teenage group included those who are currently enrolled in school and have not completed his/her education.

during their field work in the early 1990s to early 2000s (Natsuhara and Ohtsuka, 1999; Umezaki and Ohtsuka, 2002), or by using a local event calendar. Height was measured to the nearest 1 mm using a field anthropometer (GPM, Switzerland). Weight measurements were made to the nearest 0.1 kg using a digital scale (Tanita, Japan) without shoes and in light clothing. BMI was calculated as body weight (kg) divided by the square of height (m), and was rounded to one decimal place. Statistical analyses First, we performed correlation analyses to investigate the strength and direction of the association between protein intake estimate from the FFQ and the concentrations of biomarkers. Partial correlation coefficients were used to assess the association between usual total protein intake estimates from the FFQ and the first morning void UN concentration after adjusting for the influence of urinary creatinine concentration (Barr et al., 2005). Spearman’s correlation coefficients were used to assess the association between usual animal protein intake estimates from the FFQ and the hair d15N. We constructed Bland–Altman plots (Bland and Altman, 1986) to investigate the level of agreement and the presence of proportional bias between the reported protein intake from the FFQ and the corresponding biomarkers, after transformation to Z scores. With respect to usual total protein estimates from the FFQ and the UN concentration, we calculated the residuals from the regression between the FFQ-estimated protein and urinary creatinine concentration and the residuals from the regression between UN concentration and urinary creatinine concentration before transformation to Z scores. Joint classification analyses were conducted to investigate the degrees of misclassification error (Willett et al., 1985). Usual protein intake estimates from the FFQ and the biomarkers were divided into quartiles and the percentage of agreement between the methods was computed. All of the analyses were performed separately for males and females. We also compared the median daily protein intake estimated from FFQ- and 3d-WFR, using Wilcoxon signedAmerican Journal of Human Biology

rank test, to examine the validity of the FFQ to estimate quantity of usual protein intake at a group level. Median daily protein intake estimated from FFQ was then compared with the biological requirement proposed by WHO/ FAO/UNU (2007). Digestibility and protein quality of the PNG Highlanders’ mixed diets were considered. The digestibility was assumed to be 0.955 based on an earlier study conducted in the PNG Highlands (Fujita et al., 1986). Protein quality was calculated by % contribution of each food item to the community’s protein intake and the corresponding amino acid score (Japanese Ministry of Education, Culture, Sports, Science and Technology, 2005). RESULTS Characteristics of study participants In total, 135 PNG Highlanders completed the FFQ; of whom 26 also completed the WFR without a ceremonial feast (Table 1). All of the WFR participants were from Frigano. Among the FFQ participants, the proportion of males was 54.1%, and the median and range of age was 29 (12–67) years for males and 27.5 (13–66) years for females. Most were subsistence farmers and half of the participants had not received any formal education. Mean BMI was above 18.5 and mean MUAC was over 23 cm in both sexes. Among the WFR participants the proportion of males was 50.0%, and the median and range of age was 25 (12–65) years for males and 28 (14–66) years for females. The demographic characteristics of the FFQ and WFR participants were not significantly different except for formal education, which was higher for the WFR participants (P < 0.05). In both studies, male participants were taller and heavier than female participants, but they did not differ significantly by age, BMI, MUAC, years of formal schooling, or residence. FFQ validation against protein biomarkers Of the 135 participants, 129 non-pregnant participants provided urine and 126 participants provided hair samples. The median UN and urinary creatinine concentrations

353

PROTEIN INTAKE IN PAPUA NEW GUINEAN HIGHLANDERS

were 325.4 mg/dl (inter-quartile range [IQR], 210.5–604.6) and 126.7 mg/dl (IQR, 70.1–199.0), respectively. The median hair d15N was 7.7 & (IQR, 6.8–8.7). No significant gender difference was observed in the first morning void UN concentration and hair d15N but males showed significantly greater median urinary creatinine concentrations than females (151.1 mg/dl vs. 93.3 mg/dl, P < 0.05). The FFQ-estimated protein intake per day significantly correlated with UN concentration in the first morning void after adjusting for the influence of urinary creatinine concentration (partial correlation r 5 0.28, P < 0.01). When stratified by gender, the positive association remained significant in males (partial correlation r 5 0.31, P < 0.01) and there was a positive trend in females (partial correlation r 5 0.24, P < 0.10). The FFQ-estimated animal protein intake per day significantly correlated with hair d15N (Spearman’s r 5 0.34, P < 0.001). The positive association remained significant when the sample was restricted to males (Spearman’s r 5 0.41, P < 0.001) as well as to females (Spearman’s r 5 0.26, P < 0.05). Figure 1a, b display the Bland–Altman plots for Z-score transformed value of total protein and animal protein intake estimated from the FFQ and the biomarkers. Limits of agreement were between –2.39 and 2.39 (in Z score) for total protein intake and –2.19 and 2.19 for animal protein intake. The intra-individual variation increased as total and animal protein intake increased. When the participants were classified into four quartiles, based on total protein intake estimates from the FFQ and biomarkers, 27.2% showed exact matching of the classification quartile and 38.8% showed adjacent quartile matching. Gross misclassification (i.e. the first quartile was classified as the fourth quartile or vice versa) was 7.0%. When the participants were classified into four quartiles, based on animal protein intake estimates from the FFQ and biomarkers, 34.8% showed exact matching of the classification quartile and 38.8% showed adjacent quartile matching. Gross misclassification was 6.4%. FFQ validation against 3d-WFR The median daily total protein intake estimated from the FFQ was 44.5 g (IQR, 32.2–56.5) while the median obtained from the 3d-WFR was 45.5 g (IQR, 32.2–64.1). Similarly, the median daily animal protein intake estimated by the FFQ was 9.9 g (IQR, 5.3–15.6), while the median obtained from the 3d-WFR was 5.9 g (IQR, 0.0–17.0). The FFQ-estimated and 3d-WFR-obtained median protein intake did not differ significantly in total protein intake (P 5 0.56) or animal protein intake (P 5 0.33). Evaluation of habitual protein intake in PNG Highlanders Dietary intake is determined by energy requirements that greatly vary by age during the growing phase. Therefore, we hereby present FFQ-estimated total and animal protein intakes among adults (aged 20 year and over) only. Adult males consumed a median total protein of 36.0 g (IQR, 28.1-52.7) and animal protein of 5.4 g (IQR, 2.19.8) per day. Adult females consumed 31.3 g (IQR, 24.447.5) of total protein and 6.1 g (IQR, 2.0-9.6) of animal protein per day. Table 2 presents FFQ-estimated protein intake by food item among the male and female adult participants in each community. Protein intake was higher in males than females across the communities, and the

median and interquartile range of protein intake was higher in communities with better access to an urban center (Frigano > Wenani > Levani). More specifically, the contribution of root crops was reduced; and the contribution of store-bought meat and vegetables, and other subsistence vegetables (e.g. pumpkin, beans) was higher. The contribution of fast food, such as chips, remained at almost zero across the villages. The protein intake varied both between and within communities, but the withincommunity difference was larger among communities with better access to store foods. The biologically required protein intake for adults proposed by WHO/FAO/UNU (2007) is 0.66 g/kg/day. The amino acid score of PNG Highlanders’ mixed dietary protein was estimated as 0.74 (Table 3). Correcting for quality and digestibility, we estimated the biological requirement of protein intake in PNG Highlanders to be 0.93 g/kg/day (i.e. 0.66 divided by 0.74 and 0.955). The median total protein intake per kg body weight among adult male and female were 0.77 (IQR: 0.57–1.18) and 0.76 (IQR: 0.47–0.99) in Frigano, 0.66 (IQR: 0.43–0.96) and 0.88 (IQR: 0.58–0.96) in Wenani, and 0.55 (IQR: 0.43–0.61) and 0.47 (IQR: 0.37–0.65) in Levani. DISCUSSION We described the development of a 32-item FFQ to measure habitual protein intake of PNG Highlanders and examined its validity against UN concentration in the first morning void, d15N in scalp hair, and 3d-WFR data. Daily total and animal protein intake, estimated from the FFQ, correlated significantly with UN in the first morning void and hair d15N, respectively, and about 70% of the participants were classified in the same or adjacent quartiles. With respect to absolute protein intake estimates, daily total and animal protein intake estimates from the FFQ showed good agreement with the median protein intake from the 3d-WFR. We estimated the amino acid score of PNG Highlanders’ diets to be 0.74 from the FFQ data, and found that the average habitual protein intake in all the communities was below the biological requirement proposed by WHO/FAO/UNU (2007). This was especially pronounced in the village which is more distant from an urban center. This is the first reported study to develop a valid FFQ for use in a PNG Highlands population. Many FFQs have been developed for various populations globally but development of a FFQ is generally seen to be difficult in developing countries, or even in rural areas in developed countries. A review of PubMed, Web of Science, and Google Scholar (first 20 articles) for literature with the following key words in the title: (Validity OR Validation) AND (FFQ) AND (Country name) revealed that FFQs had been developed and validated in only 10 of 109 low- or middleincome countries (World Bank definition: 2012 GNI capital $12,615 or less). It has been suggested that in such countries, low literacy and food sharing practices impeded the development of FFQs (Ferro-Luzzi, 2002; Kuzma and Lidsted, 1990). We therefore designed the FFQ carefully to reduce estimation errors in a low-literacy population with food sharing practices. First, we reduced the number of questions by focusing on food items that are major protein sources in the PNG Highlands rather than targeting the entire diet. Secondly, we designed the FFQ to estimate a regular serving size for commonly shared food and American Journal of Human Biology

354

A. MORITA ET AL.

Fig. 1. Bland–Altman plot to describe agreement between the FFQ and biomarkers in Z score values (a) describes agreement for total protein intake estimates between the FFQ and UN without the influence of CRE; (b) describes for agreement for animal protein intake estimates between the FFQ and hair 15N; Tpro, total protein intake; CRE, creatinine

adjusted by body sizes of males and females. Thirdly, we conducted face-to-face interviews with trained interviewers using realistic food models. We conducted the American Journal of Human Biology

FFQ first and assessed WFR later, and used overnight UN concentration and hair d15N that do not involve recall bias as references.

355

PROTEIN INTAKE IN PAPUA NEW GUINEAN HIGHLANDERS TABLE 2. Median (IQR) protein intake of adult male and female highlanders in each community determined using the FFQ data Male

Female

Frigano (n 5 13)

Wenani (n 5 15)

Levani (n 5 30)

Frigano (n 5 16)

Wenani (n 5 13)

Levani (n 5 20)

Total protein 58.4 (34.3 - 70.6) Animal sources of protein 8.2 (4.5 - 18.2) Vegetable sources of protein 36.5 (29.1 - 62.2) Protein intake from meats, poultry and fish Subsistence foods Pork 0.0 (0.0 - 1.3) Game meat 0.0 (0.0 - 0.1) Freshwater fish 0.0 (0.0 - 0.3) Processed, store-bought foods a 2.0 (1.3–4.7) Chicken 0.7 (0.7 - 2.5) Lamb chopa Canned meat 0.0 (0.0 - 1.1) Saltwater fish 1.2 (0.1 - 1.7) Canned fish 1.8 (0.4 - 3.7) Egg 0.4 (0.0 - 1.5) b 1.2 (0.6 - 1.7) Other processed meat and fish Protein intake from vegetables and fruits Subsistence foods 8.0 (4.3 - 13.2) Root cropsc Corn 0.6 (0.4 - 2.1) Green leaves 5.1 (2.8 - 7.0) 1.0 (0.3 - 2.4) Other vegetablesd Nuts 3.9 (0.5 - 11.6) Sweet banana 0.2 (0.1 - 0.4) Processed, store-bought foods e 6.1 (2.2 - 15.7) Rice 4.8 (1.1 - 9.8) Bakeryf Noodle 1.4 (0.4 - 2.8) g 1.3 (0.6 - 2.6) Other

39.9 (25.2 - 59.9) 6.7 (2.0 - 19.8) 31.8 (23.5 - 40.1)

30.7 (27.5 - 38.8) 4.0 (1.5 - 7.7) 25.7 (23.2 - 30.7)

40.4 (27.4 - 54.0) 8.8 (4.8 - 14.2) 31.3 (21.9 - 40.1)

42.0 (27.6 - 58.6) 6.0 (2.0 - 9.3) 38.2 (22.6 - 49.0)

26.8 (20.9 - 34.0) 5.1 (1.4 - 8.9) 21.0 (15.2 - 26.8)

0.8 (0.0 - 0.8) 0.0 (0.0 - 0.0) 0.1 (0.0 - 0.6)

1.7 (0.8 - 2.7) 0.0 (0.0 - 0.0) 0.0 (0.0 - 2.5)

0.9 (0.0 - 0.9) 0.0 (0.0 - 0.3) 0.0 (0.0 - 0.1)

0.0 (0.0 - 0.9) 0.0 (0.0 - 0.0) 0.0 (0.0 - 0.5)

1.6 (0.8 - 4.1) 0.0 (0.0 - 0.0) 0.4 (0.0 - 2.7)

0.9 (0.9 - 3.8) 0.0 (0.0 - 0.0) 0.4 (0.0 - 1.5) 0.0 (0.0 - 0.0) 0.4 (0.3 - 3.5) 0.0 (0.0 - 0.6) 0.0 (0.0 - 0.0)

0.0 (0.0 - 0.0) 0.0 (0.0 - 0.0) 0.0 (0.0 - 0.0) 0.0 (0.0 - 0.0) 0.6 (0.3 - 1.2) 0.0 (0.0 - 0.0) 0.0 (0.0 - 0.0)

2.3 (0.7 - 3.1) 0.5 (0.0 - 1.1) 0.0 (0.0 - 0.5) 1.5 (0.0 - 6.2) 2.0 (0.4 - 2.7) 0.0 (0.0 - 1.2) 0.8 (0.2 - 1.7)

0.7 (0.3- 2.0) 0.0 (0.0 - 0.3) 0.2 (0.0 - 2.6) 0.0 (0.0 - 0.0) 0.4 (0.1 - 2.1) 0.0 (0.0 - 2.1) 0.0 (0.0 - 0.2)

0.0 (0.0 - 0.0) 0.0 (0.0 - 0.0) 0.0 (0.0 - 0.0) 0.0 (0.0 - 0.0) 0.2 (0.0 - 0.9) 0.0 (0.0 - 0.0) 0.0 (0.0 - 0.0)

15.4 (12.7 - 17.3) 0.0 (0.0 - 0.1) 3.3 (1.5 - 5.1) 0.7 (0.1 - 3.0) 0.7 (0.3 - 1.2) 0.4 (0.1 - 1.0)

16.0 (13.5 - 19.7) 0.0 (0.0 - 0.1) 1.7 (1.2 - 4.7) 0.8 (0.1 - 1.9) 0.5 (0.3 - 1.8) 0.0 (0.0 - 0.0)

5.4 (4.0 - 7.2) 1.0 (0.5 - 1.4) 4.0 (2.3 - 6.0) 1.8 (0.9 - 3.6) 3.4 (1.4 - 8.9) 0.2 (0.2 - 0.4)

15.3 (13.6 - 19.0) 0.1 (0.0 - 0.2) 2.8 (2.2 - 3.1) 0.6 (0.1 - 2.2) 2.6 (0.4 - 2.6) 0.2 (0.0 - 0.8)

12.3 (9.8 - 15.4) 0.0 (0.0 - 0.0) 1.3 (0.3 - 3.4) 1.3 (0.4 - 1.9) 1.5 (0.7 - 4.8) 0.0 (0.0 - 0.0)

3.0 (0.7 - 7.4) 1.9 (0.8 - 9.2) 0.5 (0.4 - 0.9) 0.7 (0.1 - 3.0)

2.9 (1.4 - 5.0) 0.0 (0.0 - 0.0) 0.4 (0.2 - 0.7) 0.8 (0.1 - 1.9)

4.4 (1.3 - 7.8) 4.5 (1.8 - 10.2) 0.7 (0.1 - 1.2) 2.2 (1.3 - 6.0)

3.5 (1.3 - 6.1) 3.1 (1.0 - 6.4) 1.0 (0.2 - 2.0) 0.6 (0.1 - 2.2)

1.8 (0.6 - 3.1) 0.0 (0.0 - 0.2) 0.2 (0.0 - 0.5) 1.3 (0.4 - 1.9)

a

Chicken and lamb chop included both home-cooked and restaurant-served meat Other processed meat and fish include wanmaus, stew, fried pork, fish sausage, and fish flour from a store. Root crops include sweet potatos, taro, yam, cassava, potato, and house beans. d Other vegetables include pumpkin and beans. e Rice includes riceballs, plain and fried rice from the store. f Bakery includes buns, bread, scorn, crackers, and donuts. g [Other] includes chips and vegetable side dishes at restaurants. b c

The present FFQ showed a reasonable agreement with protein biomarkers to rank individuals based on their protein intake. FFQ validation studies in developing countries have reported correlation coefficients between 0.30 and 0.70 against WFRs (Bautista et al., 2005; Bowen et al., 2012; Fornes et al., 2003; Hong et al., 2010; Iqbal et al., 2009; Maclntyre et al., 2001; Nurul-Fadhilah et al., 2012; Parr et al., 2002; Rodriquez et al., 2002; Slater et al., 2003; Torheim et al., 2001; Upreti et al., 2012; Xu et al., 2004). The proportion of correct classification was comparable to other FFQ outcomes (i.e. 30–40%) (Bautista et al., 2005; Bowen et al., 2012; Fornes et al., 2003; Hong et al., 2010; Iqbal et al., 2009; Maclntyre et al., 2001; Nurul-Fadhilah et al., 2012; Parr et al., 2002; Rodriquez et al., 2002; Slater et al., 2003; Torheim et al., 2001; Upreti et al., 2012; Xu et al., 2004). To date, only one study has investigated the validity of the FFQ against a biomarker in a developing country, and it failed to show a correlation (Maclyntyre et al., 2001). Thus, we can say that our FFQ survey compares favorably with FFQs designed for use in low-income settings. With respect to estimating absolute protein intake, Bland–Altman plots indicated that the difference between the FFQ-estimates and corresponding biomarkers varied over a wide range at the individual level. In general, there was a tendency for over- or under-estimation with increasing protein intake. At the group-level, however, the FFQestimates and 3d-WFR showed good agreement; the difference was less than 5 g. The protein intake varied both between and within communities but the within-community

difference was larger among communities with better access to an urban center. In Levani, for example, where people need to spend half a day to get to the nearest commercial facilities, the IQR of total protein intake was between 27.5 and 38.8 g/day. In comparison, in Frigano where people had access to the town by public transportation, the IQR of total protein intake was between 29.4 and 68.9 g per day. BMI and MUAC were well above the international cutoff points for poor nutritional status. Earlier nutrition studies indicate that protein intake in the PNG Highlands has increased gradually over the last half-century. For example, Harvey and Heywood (1983) documented that young adult males consumed greater total protein on average in the village of Yobakogl in Simbu Province in 1981 compared with 1956 and 1975, based on 6d-WFR. The median daily total protein intake was between 20 and 30 g in their 1956 and 1975 studies but was greater (40 g) in 1981. In this study, we estimated the median daily protein intake in Frigano (the community with the best access to an urban center in our study) to be 58 g. Despite the changes in food consumption habits, choices that increase protein intake, and adequate BMI and MUAC status, our FFQ-estimated habitual protein intake showed that PNG adult Highlanders still consume considerably less protein than the biological requirement proposed by FAO/WHO/UNU (2007). Discrepancy between a protein deficient diet and apparently adequate nutritional status in PNG Highlanders, the so-called low-protein adaptation, has attracted scientific attention since the 1960s (e.g., Bergersen and American Journal of Human Biology

356

A. MORITA ET AL. TABLE 3. Estimation of Adult Papua New Guinea Highlanders’ Mixed Dietary Amino Acid Score (AAS)

Food item Sweetpotato Taro, Potato, Yam, Tapioca Cooking banana Dark greens Light greens Bean Pampkin Chicken Pork Egg Lamb Wanmouse Sausage Fish Wild bird Othermeat Canned meat Canned fish Rice ball Rice Bun Bread Scorn Cracker Donut Noodle Corn Sweet banana Peanuts Brown pandanus Fastfoodd

Supply%a

AASb

Supply% adjusted AAS

Referencec

33.0 0.8 0.7 7.3 1.5 3.1 1.0 3.5 3.5 0.8 1.0 0.1 1.0 3.3 0.1 0.0 1.7 4.2 1.1 9.2 2.3 0.1 4.9 1.5 0.2 2.1 1.6 0.7 3.6 4.7 1.3

88 68 66 50 50 68 68 100 68 100 100 68 100 100 100 100 97 100 65 65 44 44 44 29 44 35 74 66 62 62 84 Total

29.0 0.6 0.5 3.7 0.8 2.1 0.7 3.5 2.4 0.8 1.0 0.1 1.0 3.3 0.1 0.0 1.6 4.2 0.7 6.0 1.0 0.0 2.1 0.4 0.1 0.7 1.2 0.5 2.2 2.9 1.1 74.3

Sweet potato, raw (2.5.a) Average [Taro, raw (2.5.a) 1 Potato, raw (2.8.a) 1 Yam, raw (2.11.a)] Banana, raw (13.64) Spinach, raw (12.117.a) Cabbage, raw (12.24.a) Green peace, raw (12.10.a) Western pumpkin, raw (12.18.a) Skinless chicken breast (9.48.b) Pork loin fat (9.77.c) Whole egg of chicken (10.5.a) Lamb without fat (9.92.b) Pork loin fat (9.77.c) Fish sausage (8.258) Carp, raw (8.139.a) Skinless chicken breast (9.48.b) Skinless chicken breast (9.48.b) Ground beef (9.22) Mackerel, raw (8.84.a) White polished rice (1.41.d) White polished rice (1.41.d) Sliced bread (1.13.a) Sliced bread (1.13.a) Sliced bread (1.13.a) Biscuit, hard (4.65.a) Sliced bread (1.13.a) Instant chinese noodle, dried (1.31.c) Corn, raw (12.83.a) Banana, raw (13.64) Peanuts, dried, raw (6.25.a) Peanuts, dried, raw (6.25.a) Average [Potato, raw (2.8.a) 1 Skinless chicken breast (9.48.b)]

These data were derived from 121 participants who completed the FFQ and provided both urine and hair samples. a Supply%: Contribution of each food item to total protein intake calculated by [Total habitual protein intake derived from a specific food item]/[Total habitual protein intake (g/d) among the subjects]*100. b AAS of each food item was refererred to that of the same or similar food item on Standard Tables of Food Composition in Japan (1982). c Name of food items and the reference number in Standard Tables of Food Composition in Japan (1982) are shown. d For fastfood, average AAS of raw potato and skinless chicken breast was applied as chips and friedchicken accounted for 40% of the reported fastfood.

Hipsley, 1970; Koishi, 1990; Luyken et al, 1964). Nitrogen fixation and urea nitrogen salvage in the intestine have been suggested as potential mechanisms that explain the above discrepancy; however, no solid evidence exists to support these theories. The FFQ reported in the present study will enable us to estimate individual level protein intake in PNG Highlanders, which will contribute to further nutritional studies of “low protein adaptation.” Minor limitations in the study design warrant discussion. To date, we have been unable to assess the reproducibility of the FFQ. Due to the small sample size of WFR participants, we have not been able to investigate the relative validity of the FFQ’s quantity estimation among other sub-groups categorized by educational level and BMI. Marks et al. (2006) showed that in an evaluation of FFQ in comparison with WFR, age, education, and BMI played roles in estimation errors. Future studies of the validity of our FFQ in subgroups would provide information useful for improving the estimation accuracy in the PNG Highlands. CONCLUSIONS The 32-item FFQ-estimated protein intake showed a reasonable agreement with biomarkers to rank individuals based on protein intake and a good agreement with 3d-WFR-estimated protein intake. The FFQ is a valuable tool for evaluating protein intake of PNG Highlanders whose low protein adaptation mechanisms have attracted the interest of human biologists. American Journal of Human Biology

ACKNOWLEDGEMENTS This research was made possible with the commitment and cooperation we received from all of the participants and their communities. We are grateful to all of the participants and their communities for their commitment and cooperation which made this study possible. We also express our thanks to Dr. Yuichi Naito (the University of Tokyo, Japan and University of Tuebingen, Germany) and Mr. Takumi Tsutaya (the University of Tokyo, Japan) for analyzing the hair samples. LITERATURE CITED Bailey KV, J Whiteman. 1963. Dietary Studies in the Chimbu (New Guinea Highlands). Trop Geogr Med 15:377–388. Barr DB, Wilder LC, Caudill SP, Gonzalez AJ, Needham LL, Pirkle JL. 2005. Urinary creatinine concentrations in the US population: Implications for urinary biologic monitoring measurements. Environ Health Perspect 113:192–200. Bautista LE, Herran OF, Pryer JA. 2005. Development and simulated validation of a food-frequency questionnaire for the Colombian population. Public Health Nutr 8:181–188. Bergersen FJ, Hipsley EH. 1970. The presence of N2-fixing bacteria in the intestines of man and animals. J Gen Microbiol 60:61–65. Bingham SA. 2003. Urine nitrogen as a biomarker for the validation of dietary protein intake. J Nutr 133:921–924. Bingham SA, Cummings JH. 1985. Urine nitrogen as an independent validatory measure of dietary-intake—a study of nitrogen-balance in individuals consuming their normal diet. Am J Clin Nutr 42:1276– 1289. Bland JM, Altman DG. 1986. Statistical methods for assessing agreement between two methods of clinical measurement. The Lancet 327:307–310.

PROTEIN INTAKE IN PAPUA NEW GUINEAN HIGHLANDERS Bol R, Pflieger C. 2002. Stable isotope (C-13, N-15 and S-34) analysis of the hair of modern humans and their domestic animals. Rapid Commun Mass Spectrom 16:2195–2200. Bourke RM, Allen B. 2010. Nutrition surveys. In: Bourke RM, Harwood T, editors. Food and agriculture in Papua New Guinea. Canberra: ANU E Press. p 464–467. Bowen L, Bharathi AV, Kinra S, DeStavola B, Ness A, Ebrahim S. 2012. Development and evaluation of a semi-quantitative food frequency questionnaire for use in urban and rural India. Asia Pac J Clin Nutr 21:355–360. Chen Y, Ahsan H, Parvez F, Howe GR. 2004. Validity of a food-frequency questionnaire for a large prospective cohort study in Bangladesh. Br J Nutr 92:851–859. Dignan CBB, Kumar S, Aalbersberg W. 2004. The Pacific Islands food composition tables. Rome: FAO. 135 p. Ferro-Luzzi A. 2002. Keynote paper: individual food intake survey methods. Measurement and assessment of food deprivation and undernutrition. Rome: FAO. Food Standards Australia New Zealand. 2008. Australian food, supplement & nutrient database 2007. Fornes NS, Stringhini MLF, Elias BM. 2003. Reproducibility and validity of a food-frequency questionnaire for use among low-income Brazilian workers. Public Health Nutr 6:821–827. Fujita Y, Okuda T, Rikimaru T, Ichikawa M, Miyatani S, Kajiwara NM, Date C, Minamide T, Koishi H, Alpers MP, Heywood PF. 1986. Studies of nitrogen balance in male Highlanders in Papua New Guinea. J Nutr 116:536–544. Fuller BT, Fuller JL, Sage NE, Harris DA, O’Connell TC, Hedges REM. 2005. Nitrogen balance and delta N-15, why you’re not what you eat during nutritional stress. Rapid Commun Mass Spectrom 19:2497–2506. Gibson J, Rozelle S. 1998. Results of the household survey component of the 1996 poverty assessment for Papua New Guinea. A research report for the human resources department, Papua New Guinea and Pacific Islands Division: The World Bank. Harvey PW, Heywood PF. 1983. Twenty-five years of dietary change in Simbu province, Papua New Guinea. Ecol Food Nutr 13:27–35. Hipsley EH, Clements FW. 1950. Report of the New Guinea nutrition survey expedition 1947. Sydney: Government Printer. 308 p. Hong TK, Dibley MJ, Sibbritt D. 2010. Validity and reliability of an FFQ for use with adolescents in Ho Chi Minh City, Vietnam. Public Health Nutr 13:368–375. Iqbal R, Ajayan K, Bharathi AV, Zhang XH, Islam S, Soman CR, Merchant AT. 2009. Refinement and validation of an FFQ developed to estimate macro- and micronutrient intakes in a south Indian population. Public Health Nutr 12:12–18. Japanese Ministry of Education, Culture, Sports, Science and Technology. 2005. Standard tables of food composition in Japan. Koishi H. 1990. Nutritional adaptation of Papua New Guinea highlanders. Eur J Clin Nutr 44:853–885. Kuzma JW, Lindsted KD. 1990. Determinants of eight-year diet recall ability. Epidemiology 1:386–391. Luyken R, Pikaar NA, Luekenko FW. 1964. Nutrition studies in New Guinea. Am J Clin Nutr 14:13–27. Loussouarn G. 2001. African hair growth parameters. Br J Dermatol 145: 294–297. Maclntyre UE, Venter CS, Vorster HH. 2001. A culture-sensitive quantitative food frequency questionnaire used in an African population. II. Relative validation by 7-day weighed records and biomarkers. Public Health Nutr 4:63–71. Marks GC, Hughes MC, van der Pols JC. 2006. Relative validity of food intake estimates using a food frequency questionnaire is associated with sex, age, and other personal characteristics. J Nutr 136:459–465 Mojtahedi M, de Groot LCPGM, Boekholt HA, van Raaij JMA. 2001. Nitrogen balance of healthy Dutch women before and during pregnancy. Am J Clin Nutr 75:1078–1083. Muntwiler M, Shelton RM. 2001. Survey of nutrition and protein intake in rural families in Eastern Highlands Province. In: Bourke RM, Allen B, Salisbury JG, editors. Food security for Papua New Guinea. Canberra: Australian Centre for International Agricultural Research. p 432–442. Natsuhara K. 2002. Nutritional ecology of modernizing rural communities in Papua New Guinea: influences of food sharing on nutritional and health status [PhD thesis]. Tokyo: The University of Tokyo. Natsuhara K, Ohtsuka R. 1999. Nutritional ecology of a modernizing rural community in Papua New Guinea: an assessment from urinalysis. Man and Culture in Oceania 15:95–115. Norgan NG, Ferroluz A, Durnin JVG. 1974. Energy and nutrient intake and energy expenditure of 204 New Guinean adults. Philos Trans R Soc Lond B Biol Sci 268:309–348.

357

Nurul-Fadhilah A, Teo PS, Foo LH. 2012. Validity and reproducibility of a food frequency questionnaire (FFQ) for dietary assessment in Malay adolescents in Malaysia. Asia Pac J Clin Nutr 21:97–103. O’Connell TC, Hedges REM. 1999. Investigations into the effect of diet on modern human hair isotopic values. Am J Phys Anthropol 108:409–425. Oomen HA. 1972. Distribution of nitrogen and composition of nitrogen compounds in food, urine and faeces in habitual consumers of sweet potato and taro. Nutr Metab 14:65–82. Parr CL, Barikmo I, Torheim LE, Ouattara F, Kaloga A, Oshaug A. 2002. Validation of the second version of a quantitative food-frequency questionnaire for use in Western Mali. Public Health Nutr 5:769–781. Petzke KJ, Boeing H, Metges CC. 2005. Choice of dietary protein of vegetarians and omnivores is reflected in their hair protein C-13 and N-15 abundance. Rapid Commun Mass Spectrom 19:1392–1400. Petzke KJ, Fuller BT, Metges CC. 2010. Advances in natural stable isotope ratio analysis of human hair to determine nutritional and metabolic status. Curr Opin Clin Nutr Metab Care 13:532–540. Reitsema LJ. 2013. Beyond diet reconstruction: Stable isotope applications to human physiology, health, and nutrition. Am J Hum Biol 25:445–456. Rodriguez MM, Mendez H, Torun B, Schroeder D, Stein AD. 2002. Validation of a semi-quantitative food-frequency questionnaire for use among adults in Guatemala. Public Health Nutr 5:691–698. Rush EC, Freitas I, Plank LD. 2009. Body size, body composition and fat distribution: comparative analysis of European, Maori, Pacific Island and Asian Indian adults. Br J Nutr 102:632–641. Sinnett PF. 1972. Nutrition in a New Guinea Highland community. Hum Biol Oceania 1:299–305. Sinnett PF, Solomon A. 1968. Physical fitness in a New Guinea Highland. P N G Med J 11:56–59. Slater B PS, Fisberg RM, Latorre MR. 2003. Validation of a semiquantitative adolescent food frequency questionnaire applied at a public school in S~ ao Paulo, Brazil. Eur J Clin Nutr 57:629–635. Tang AM, Dong K, Deitchler M, Chung M, Maalouf-Manasseh Z, Tumilowicz A, Wanke C. 2013. Use of cutoffs for mid-upper arm circumference (MUAC) as an indicator or predictor of nutritional and healthrelated outcomes in adolescents and adults: a systematic review. Washington: FANTA. p 34. Torheim LE, Barikmo I, Hatloy A, Diakite M, Solvoll K, Diarra MM, Oshaug A. 2001. Validation of a quantitative food-frequency questionnaire for use in Western Mali. Public Health Nutr 4:1267–1277. Upreti D KJ, Simkhada P, Thoulass J, McNeill G. 2012. Development and validation of a food frequency questionnaire for assessment of diet among people living with HIV/AIDS in Nepal. P Nutr Soc 71:E5. Umezaki M, Ohtsuka R. 2002. Changing migration patterns of the Huli in the Papua New Guinea highlands–A genealogical demographic analysis. Mt Res Dev 22:256–262. Umezaki M, Natsuhara K, Ohtsuka R. 2001. Protein content and amino acid scores of sweet potatoes in Papua New Guinea Highlands. Ecol Food Nutr 40:471–480. Umezaki M, Yamauchi T, Ohtsuka R. 1999. Diet among the Huli in Papua New Guinea Highlands when they were influenced by the extended rainy period. Ecol Food Nutr 37:409–427. Ulijaszek SJ, Hyndman DC, Lourie JA, Pumuye A. 1987. Mining, modernization and dietary change among the Wopkaimin of Papua-NewGuinea. Ecol Food Nutr 20:143–156. Yamauchi T. 2001. Impact of microenvironment on food security and nutritional adaptation in the Tari basin. In: Bourke RM, Allen B, Salisbury JG, editors. Food Security for Papua New Guinea. Canberra: Australian Centre for International Agricultural Research. p 432–442. Yoshinaga J, Minagawa M, Suzuki T, Ohtsuka R, Kawabe T, Hongo T, Inaoka T, Akimichi T. 1991. Carbon and nitrogen isotopic characterization for Papua-New-Guinea Foods. Ecol Food Nutr 26:17–25. Yoshinaga J, Minagawa M, Suzuki T, Ohtsuka R, Kawabe T, Inaoka T, Akimichi T. 1996. Stable carbon and nitrogen isotopic composition of diet and hair of Gidra-speaking Papuans. Am J Phys Anthropol. 100:23–34. Weiner J, Lourie JA. 1981. Practical human biology. London: Academic Press. p 439. WHO/FAO/UNU. 2007. Protein and amino acid requirements in human nutrition. World Health Organ Tech Rep Ser 935:1–265. Willett WC. 1986. Nutritional epidemiology. New York: Oxford University Press. Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, Hennekens CH, Speizer FE. 1985. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 122:51–65. Xu LZ, Dibley MJ, D’Este C. 2004. Reliability and validity of a foodfrequency questionnaire for Chinese postmenopausal women. Public Health Nutr 7:91–98.

American Journal of Human Biology

Copyright of American Journal of Human Biology is the property of John Wiley & Sons, Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.