This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Research Protocols, is properly cited. The complete bibliographic information, a link to the original publication on http://www.researchprotocols.org, as well as this copyright and license information must be included.
There is very little published work on dietary intake and nutritional status of Mozambicans. We conducted a population-based cross-sectional study on the diet and nutritional status of adolescent girls in different types of communities in Zambézia Province, Central Mozambique, in two distinct seasons.
The purpose of this paper is to present the design, methods, and study population characteristics of the Estudo do Estado Nutricional e da Dieta em Raparigas Adolescentes na Zambézia (the ZANE Study).
Data was collected in January-February 2010 ("hunger season") and in May-June 2010 ("harvest season"). A total of 551 girls in the age group 14-19 years old were recruited from one urban area and two districts (district towns and rural villages). The study protocol included a background interview, a 24-hour dietary recall interview, a food frequency questionnaire, anthropometric measurements, bioimpedance, hemoglobin measurement, and venous blood, urine, buccal cell, and fecal sampling.
Adolescent motherhood was common in all study regions. Stunting prevalence for the total study population as a weighted percentage was 17.8% (95/549; 95% CI 14.3-22.0) with no regional differences. Overweight was found mainly in the urban area where the prevalence was 12.6% (20/159; 95% CI 7.5-17.6), thinness was rare. There were regional differences in the prevalence of malaria parasitemia and intestinal helminth infestation, but not human immunodeficiency virus.
The fully analyzed data from the ZANE Study will yield results useful for setting priorities in nutrition policy and further research on the diet and nutritional status in Mozambique and other countries with similar nutritional problems.
ClinicalTrials.gov: NCT01944891; http://www.clinicaltrials.gov/ct2/show/NCT01944891 (Archived by WebCite at http://www.webcitation.org/6L9OUrsq8).
Mozambique is one of the least developed countries in the world with almost 60% of the population living below the international poverty line [
While the problem of undernutrition persists in Mozambique and much of Southern Africa, the region is also experiencing the phenomenon of nutrition transition [
There is a lack of published work on dietary intake and biochemical indicators of the nutritional status of Mozambicans, especially in age groups other than children below 5 years in age. There are two studies, one for women [
With regard to biochemical indicators of nutritional status, anemia is known to be a severe problem in Mozambique—the Demographic and Health Survey 2011 found that 54% of girls and women of 15 to 49 years of age were anemic [
In the
The Bioethical Committee of the Ministry of Health in Mozambique approved the study plan.
Zambézia Province with 3.89 million inhabitants [
Quelimane City has 196,000 inhabitants [
Map of Zambézia Province. The original maps are from d-maps.com [
Adolescent girls in the age range of 15 to 18 years old, residing in the selected study regions, constituted the target study population. The sample size of 600 girls was determined based on the resources available. A sample size of 100 girls from each region per season was considered feasible. The number of primary sampling units (PSUs) selected from each region was dependent on practical considerations. The aim was to have an equal sample size from each PSU.
For purposes of the study, the two districts were divided into district towns and rural areas. The sampling was carried out in five areas: (1) Quelimane City, (2) the district town of Maganja da Costa (
In the city and two district towns, data from the 2007 Census provided information on the population sizes of the neighborhoods. One neighborhood with a small population size was removed from the sample framework. In the two rural areas, locally obtained maps provided information on the villages. Villages located within a 45 to 60 minute drive from the study centers (health centers and hospitals) were considered as PSUs. There were no available listings of households in the PSUs.
The study was undertaken in two seasons—January-February, 2010 (“hunger season”) and May-June, 2010 (“harvest season”). The PSUs were selected using probability proportional to size (in the city and the district towns) or random sampling (in rural areas). In the city, sampling was carried out only in the catchment areas of three health centers out of the nine health centers across the city for practical reasons. The same PSUs were used in both seasons. For the second stage, the recruiters sketched a map of each selected PSU with the help of local leaders. The recruiters were instructed to follow a recruitment plan which included randomly selected starting points on the map and randomly selected directions to walk in order to chart the households. The recruiters were instructed to select households using preset intervals.
Girls in the target age range, who were able to visit the study center and did not have any significant illness preventing participation, were considered eligible. Eligible girls along with their parent, husband, or guardian were given an informed consent letter. The purpose of the study, the measurements and tests taken, the voluntary nature of participation, and the right to refuse to participate in any part of the study were also explained orally. The girl signed the informed consent form. If she was under 18 years of age, a parent, husband, or guardian also signed the form. If the girl or the adult was illiterate, a fingerprint or a signature of a witness indicated informed consent. The date of birth was verified by checking identity cards whenever possible. Alternatively, the time of birth was estimated with the help of family members. No incentives to participate in the study were given. The recruited girls were invited to come to the study center on the following day, and participants who lived far from the study center were provided with transport.
The total number of eligible girls refusing to participate is not known. Reasons for refusing to participate included refusal of parents to give permission to be part of the study, not having the time, and fear of the hospital or of blood sampling. A total of 639 girls agreed to participate, of whom 551 actually participated in the study (
A total of 143 participants were assigned to a subsample for an additional 24-hour dietary recall interview in January-February 2010 and two additional 24-hour recalls in May-June 2010. There was a total of four 24-hour recalls for the participants of this subsample. Because most of them did not have a precise address or a phone number, recontacting the participants of this subsample proved to be difficult. In January-February 2010, we were able to recontact 109 of the subsample participants for the second interview. In May-June 2010, 96 of those were successfully recontacted for a third interview, and 84 of those for a fourth interview. The additional 24-hour recalls were conducted on nonconsecutive days.
There were practical difficulties obtaining all information needed for multi-stage cluster sampling, for example, detailed maps with demarcation of all PSUs were not available and the population size of eligible girls in all PSUs was not known. Furthermore, recruiting an equal sample size from each selected PSU turned out not to be successful. In some cases, nonrandomly selected PSUs had to be included. For example, a selected village had to be replaced by another one when the leaders of the village were not supportive of the study. Taking into consideration all of these factors, sampling weights were calculated using the total population sizes of 15-19-year-old girls in the five areas from the 2007 Census. The population sizes used for calculating the sampling weights are shown (underlined) in
Implemented sampling design. The population figures are from the 2007 Census. y=years.
All forms and questionnaires used in the study were in Portuguese. However, if the participant was not comfortable in Portuguese, the questions were explained by field workers and appropriate interpreters in a language understood by the participant. The background questionnaire included questions about the characteristics of the participant and her household, for example, schooling, mother tongue, marital status, the number of children, pregnancy and breastfeeding status, self-reported health status, usage of mosquito net, cigarette and alcohol use, exposure to media (television, radio, newspaper), ownership of livestock and other assets, sanitation, and the type of housing. Asking each participant to read out a short test sentence in Portuguese tested literacy; being literate was defined as the ability to read the whole sentence. The background interview also included a Portuguese version of Household Food Insecurity Access Scale (HFIAS) questions [
A 7-day food frequency questionnaire (FFQ) was designed specifically for this study. The list of foods included in the FFQ was based on the experience accumulated during preliminary fieldwork and questionnaires on dietary diversity or food consumption from previously published [
For the 24-hour dietary recall interview, the participant was asked to recall and report on the main activities that she had engaged in during the previous day. She was then asked to report all the food and beverages she had consumed during that period. The information gathered from the participant on the daily activities was used for probing. With regard to composite dishes, the participant was asked to describe all the ingredients in the dish. For the data collection period of May-June 2010, a separate question about the use of iodized salt was added to the form.
A set of food photographs was shown to the participant to help her with the estimation of portion sizes (
To calculate nutrient intake, a Mozambican food composition database was compiled. The majority of the food composition data was taken from the United States Department of Agriculture National Nutrient Database for Standard Reference [
Recipe data was collected during visits to local households by recording the weight of each ingredient and the prepared dish. Of the collected recipes, 24 were relevant and directly useable, or could be modified by changing some ingredients. In addition, yield factors (to adjust for weight changes during cooking) for a few foods that include only one main ingredient in addition to water, for example, thick maize porridge, were collected. Some yield factors were also adapted from other sources [
Example of food photographs used in portion size estimation: 270 g, 408 g, and 610 g of thick maize porridge.
For the most part, anthropometric measurements were carried out according to the WHO technical instructions [
The participants’ waist circumference, right upper arm length, and mid-upper arm circumference were measured after marking of the anatomical sites [
The participants’ triceps and subscapular skinfold measurements were measured using a Harpenden Skinfold Caliper (Baty International, UK) according to the instructions of the manufacturer. The anatomical sites were marked [
Hand-to-foot bioimpedance (Bodystat 1500 DMM, Bodystat Ltd, UK) was used to measure body composition of the participants. Bioimpedance measurements were taken in the standing position [
A member of the field team used a digital automatic blood pressure monitor (Omron M6, Omron Healthcare, Japan) to measure the blood pressure and pulse of the participants. Only one measurement was taken as most participants were unfamiliar with the procedure and the protocol did not allow time for them to become accustomed to it.
Participants’ forearm muscle strength measurements were taken using a hydraulic hand dynamometer (SH5001, Saehan Corporation). The participant was encouraged to grip the handle with all her strength. There were two successive measurements that were taken for both hands. The result used in the study is the higher reading from each hand.
The recruiters gave participants a fecal sample container and instructed them to deposit the sample at home. The laboratory technicians performed a direct wet mount with saline and examined the specimen (n=353) for the presence of helminth eggs or larvae under a microscope.
During the study center visit, the participants were asked to provide a spot urine sample. The samples were tested for pregnancy (Insight-HCG, Tulip Diagnostics, India; or Onsite HCG Combo Rapid Test, CTK Biotech, USA) and aliquots were frozen for further analyses. Although it was not part of the study protocol, unprompted, the laboratory technicians of Morrumbala District checked suspect (samples with dark brown or red coloration) urine samples for the presence of
Buccal cells were obtained by scraping the inside of the cheek of each participant with a new toothbrush [
Venous blood samples (n=515) were taken from the antecubital vein using 21 g needles in serum tubes (10 ml) and 3 ml tubes containing potassium ethylenediaminetetraacetic acid (EDTA) (K3EDTA) (BD Vacutainer, Becton Dickinson International, Belgium). The EDTA-blood sample was centrifuged to plasma at 4500 rpm (1700 x g) for 10 minutes and aliquots were frozen and stored at -15 to -20 °C. The serum tubes were allowed to stand at room temperature (22-36 °C) for 30 minutes, and then centrifuged and frozen as described above. The study team purchased freezers to ensure adequate cold storage in the study centers. The samples were transported to Maputo in a portable freezer by car where they were packed in dry ice and shipped to Finland. In Finland, they were stored at -70 °C.
EDTA-blood was used for hemoglobin determination with HemoCue Hb 301 System (HemoCue AB, Sweden). The presence of malaria (
Drops of EDTA-blood were applied to the sample collection area of Whatman 903 Protein Saver Cards, dried, and packed in Ziploc storage bags with desiccant packs. The cards were kept at room temperature until delivered to Finland, where they were stored at -20 ˚C.
A local nurse gave the results of the pregnancy, hemoglobin, malaria, and intestinal helminth infestation tests to the participants, and provided iron supplements for those with hemoglobin levels below the local cut-off, 110 g/l. Antihelminthics and malaria treatments were also provided for those participants whose test results indicated these conditions. Participants who tested positive for HIV were referred to the physician who was responsible for the study in a particular region in order to have counseling and subsequent treatment.
Numbers and proportions of participants for selected background characteristics were computed. It should be noted that pregnancy was defined based on a urine test result or the participant being visibly pregnant. If this information was not available, those who self-reported not being pregnant were coded as not pregnant and those who self-reported being pregnant were coded as missing.
Z-scores for body mass index-for-age (BMI) and height-for-age were calculated using a macro by WHO [
Prevalence estimates with 95% confidence intervals (CI) were computed for the categories of BMI-for-age and height-for-age z-scores, HHS, HIV, malaria parasitemia, and presence of intestinal helminths. For these analyses, the data from the district towns and rural areas were aggregated in each district in order to present results that are representative at district level. Seasonal differences in each area were examined before combining the two seasons, and regional differences were examined with the seasons combined. Proportions were compared using Pearson's chi-square tests with the Rao and Scott second order correction. Prevalence estimates for the aggregated total study population were also computed for selected variables. Sampling weights, as described above, were used in the analyses. Analyses were carried out using the survey package in R version 3.0.1.
The exact date of birth was known for 75.5% (416/551) of the participants, and for 92.1% (383/416) of those, the recruiters reported having verified the date of birth from an identity card or other official documents. For 24.5% (135/551) of the participants, only the year or year and month of birth were recorded.
The background characteristics (
Characteristics of the study population.
|
Quelimane | Maganja da Costa District | Morrumbala District | |||
|
|
District town | Rural villages | District town | Rural villages | |
|
n=179 | n=57 | n=127 | n=49 | n=139 | |
|
|
|
|
|
|
|
|
14 years | 10 (5.6) | 2 (4) | 5 (3.9) | 1 (2) | 9 (6.5) |
|
15 years | 52 (29.1) | 19 (33) | 45 (35.4) | 21 (43) | 62 (44.6) |
|
16 years | 39 (21.8) | 9 (16) | 27 (21.3) | 9 (18) | 30 (21.6) |
|
17 years | 46 (25.7) | 13 (23) | 22 (17.3) | 7 (14) | 18 (12.9) |
|
18 years | 32 (17.9) | 13 (23) | 27 (21.3) | 11 (22) | 18 (12.9) |
|
19 years | 0 (0.0) | 1 (2) | 1 (0.8) | 0 (0) | 2 (1.4) |
|
Missing data, n | 0 | 0 | 0 | 0 | 0 |
|
||||||
|
n (%) | 170/177 (96.0) | 49/56 (88) | 77/127 (60.6) | 42/49 (86) | 91/134 (67.9) |
|
Missing data, n | 2 | 1 | 0 | 0 | 5 |
|
||||||
|
n (%) | 145/178 (81.5) | 28/53 (53) | 39/124 (31.5) | 20/48 (42) | 15/134 (11.2) |
|
Missing data, n | 1 | 4 | 3 | 1 | 5 |
|
||||||
|
n (%) | 10/179 (5.6) | 7/56 (13) | 30/127 (23.6) | 4/49 (8) | 40/133 (30.1) |
|
Missing data, n | 0 | 1 | 0 | 0 | 6 |
|
||||||
|
n (%) | 18/178 (10.1) | 6/56 (11) | 20/125 (16.0) | 1/49 (2) | 15/135 (11.1) |
|
Missing data, n | 1 | 1 | 2 | 0 | 4 |
|
||||||
|
n (%) | 38/176 (21.6) | 16/55 (29) | 38/125 (30.4) | 10/49 (20) | 41/133 (30.8) |
|
Missing data, n | 3 | 2 | 2 | 0 | 6 |
|
||||||
|
n (%) | 9/177 (5.1) | 8/55 (15) | 9/127 (7.1) | 6/49 (12) | 23/135 (17.0) |
|
Missing data, n | 2 | 2 | 0 | 0 | 4 |
|
|
|
|
|
|
|
|
No toilet | 16/179 (8.9) | 2/56 (4) | 54/127 (42.5) | 2/49 (4) | 79/135 (58.5) |
|
Pit latrine | 141/179 (78.8) | 53/56 (95) | 72/127 (56.7) | 47/49 (96) | 56/135 (41.5) |
|
Flush toilet | 22/179 (12.3) | 1/56 (2) | 1/127 (0.8) | 0/49 (0) | 0/135 (0.0) |
|
Missing data, n | 0 | 1 | 0 | 0 | 4 |
|
|
|
|
|
|
|
|
Firewood | 2/178 (1.1) | 28/56 (50) | 114/127 (89.8) | 19/49 (39) | 121/136 (89.0) |
|
Charcoal or coal | 175/178 (98.3) | 27/56 (48) | 13/127 (10.2) | 30/49 (61) | 15/136 (11.0) |
|
Electricity or gas | 1/178 (0.6) | 1/56 (2) | 0/127 (0.0) | 0/49 (0) | 0/136 (0.0) |
|
Missing data, n | 1 | 1 | 0 | 0 | 3 |
aable to read a full test-sentence in Portuguese
bmarried or traditionally married, includes divorced or separated and widowed
cpositive—urine test result positive/visibly pregnant; negative—urine test result negative, or if test result not available, self-report of not being pregnant
The majority of the girls had a BMI-for-age in the normal range and overweight was found mainly in the city (
There were eleven participants that were classified as living in households suffering from severe hunger according to the HHS. The two categories “moderate” and “severe household hunger” were combined for analysis. Household hunger prevalence estimate for the total study population as a weighted percentage was 17.0% (101/541; 95% CI 13.6-21.1) with no statistically significant regional differences.
Prevalence estimates (weighted percentages) for categories of height-for-age and BMI-for-age z-scores, and Household Hunger Scale in the study regionsa.
|
Quelimane, n=179 | Maganja da Costa District, n=184 | Morrumbala District, n=188 |
|
|
|
.329 | ||||
|
normal (≥ -2 SD), n (weighted %) | 156/178 (87.6) | 149/183 (83.0) | 149/188 (80.2) |
|
|
95% CI | 82.9-92.4 | 76.9-89.0 | 74.1-86.3 |
|
|
stunted (< -2 SD), n (weighted %) | 22/178 (12.4) | 34/183 (17.0) | 39/188 (19.8) |
|
|
95% CI | 7.6-17.1 | 11.0-23.1 | 13.7-25.9 |
|
|
Missing data, n | 1 | 1 | 0 |
|
|
< .001 | ||||
|
normal (-2 to 1 SD), n (weighted %) | 136/159 (85.5) | 150/154 (98.7) | 159/168 (95.5) |
|
|
95% CI | 80.2-90.9 | 96.9-100 | 92.1-98.8 |
|
|
thin or severely thin (< -2 SD), n (weighted %) | 3/159 (1.9) | 2/154 (1.0) | 6/168 (4.0) |
|
|
95% CI | 0-4.0 | 0-2.8 | 0.7-7.3 |
|
|
overweight (> 1 SD), n (weighted %) | 20/159 (12.6) | 2/154 (0.3) | 3/168 (0.5) |
|
|
95% CI | 7.5-17.6 | 0-0.7 | 0-1.1 |
|
|
Missing data or pregnant, n | 20 | 30 | 20 |
|
|
.274 | ||||
|
little or no HH, n (weighted %) | 137/177 (77.4) | 148/179 (81.9) | 155/185 (85.4) |
|
|
95% CI | 71.4-83.4 | 75.5-88.2 | 80.0-90.8 |
|
|
moderate or severe HH, n (weighted %) | 40/177 (22.6) | 31/179 (18.1) | 30/185 (14.6) |
|
|
95% CI | 16.6-28.6 | 11.8-24.5 | 9.2-20.1 |
|
|
Missing data, n | 2 | 5 | 3 |
|
aSampling weights were used. Seasonal differences within each area were tested with chi-square test before pooling data from two seasons and significant differences were found for HHS in Quelimane (Jan-Feb: weighted percentage 30%, 27/90, vs May-June: weighted percentage 15%, 13/87;
bOverall test (chi-square) of regional differences after pooling data from the two seasons. This test was followed by pairwise tests.
cIn pairwise comparisons, Quelimane differed significantly from Maganja da Costa (
The prevalence estimate for HIV among the total study population as a weighted percentage was 6.0% (29/372; 95% CI 3.8-9.3); there were no differences between regions (
The prevalence estimate for intestinal helminth infestation among the total study population as a weighted percentage was 11.7% (48/353; 95% CI 8.4-16.1). Significantly higher prevalence of intestinal helminths was found for Quelimane as compared to the districts (
Numbers of positive cases and prevalence estimates (weighted percentages) for HIV, malaria parasitemia, and intestinal helminth infestation in the study regionsa.
|
Quelimane, n=179 | Maganja da Costa District, n=184 | Morrumbala District, n=188 |
|
|
|
|||||
|
n (weighted %) | 10/103 (9.7) | 10/117 (7.4) | 9/152 (4.5) | .32 |
|
95% CI | 4.1-15.3 | 2.0-12.7 | 1.2-7.9 |
|
|
Missing data, n | 76 | 67 | 36 |
|
|
|||||
|
n (weighted %) | 2/164 (1.2) | 42/168 (26.2) | 2/178 (0.9) | < .001 |
|
95% CI | 0-2.9 | 18.6-33.7 | 0-2.3 |
|
|
Missing data, n | 15 | 16 | 10 |
|
|
|||||
|
n (weighted %) | 28/105 (26.7) | 7/138 (6.8) | 13/110 (13.1) | .007 |
|
95% CI | 18.4-35.0 | 2.0-11.7 | 6.4-19.9 |
|
|
Missing data, n | 74 | 46 | 78 |
|
aSampling weights were used. Seasonal differences within each area were tested with chi-square test before pooling data from two seasons and significant differences were found for intestinal helminths in Maganja da Costa (Jan-Feb: weighted percentage 12%, 7/71, vs May-June: weighted percentage 0%, 0/67;
bOverall test, chi-square, of regional differences after pooling data from the two seasons. This test was followed by pairwise tests.
cIn pairwise comparisons, Quelimane differed significantly from Maganja da Costa (
dIn pairwise comparisons, Quelimane differed significantly from Maganja da Costa (
eThe species identified:
We have collected data that will provide in-depth information on the nutritional situation of adolescent girls in Zambézia Province, Central Mozambique. The purpose of this paper is to describe the design and methods of the ZANE Study and to present descriptive results on the characteristics of the study population.
The proportion of adolescent girls living in households suffering from at least a moderate level of hunger is markedly lower in our study compared to the 57% [
Although thinness was found to be rare, close to one in five of the adolescent girls in our study area were stunted. Previously, a school-based study in the capital city Maputo found that of 2.3% of girls 6 to 18 years of age were stunted [
We found that overweight was more common in the city compared to the districts. Although this result is based on cross-sectional data, it could be interpreted as an early sign of nutrition transition in the urban adolescent girls in Central Mozambique. The finding also supports nationally representative data on adults [
The estimate for HIV prevalence was similar to the national prevalence among 15-19-year-old girls, 9.3% [
Challenges were encountered during the fieldwork. The lack of detailed maps made it difficult to plan the sampling beforehand, and the sampling design had to be adjusted during the fieldwork. Although the field team had strong local representation through the inclusion of district agricultural and health workers, in a few cases building trust with the local community leaders was not successful, and for that reason not all villages initially selected in the sampling were included in the study. The timetable for carrying out the sampling was restricted, and when recruitment was not possible at a chosen location, a village had to be chosen nonrandomly as a replacement in order to ensure the continuity of the daily flow of participants throughout both study periods. Despite these limitations in the sampling design, recruitment of participants was largely successful, and the data provides a good representation of the target population in the areas studied.
Informed consent is a critical component of study ethics. It has been shown in other studies that participants may have difficulties understanding all aspects of the informed consent; this applies for both low-income country and high-income country settings [
To summarize, the results on the characteristics of the study population of the ZANE Study indicate that the girls suffer from deprivation; the regions studied are characterized by poor sanitation facilities, and adolescent girls suffer under the burden of household hunger, infectious diseases, and intestinal helminths. The rate of adolescent motherhood is high in all the study areas, and literacy rates are particularly low in the rural areas. The forthcoming results on dietary intake and nutritional status will be useful for setting priorities in nutrition policy, and for further research on adolescent girls in Mozambique or countries with similar nutritional problems.
body mass index
confidence interval
ethylenediaminetetraacetic acid
food frequency questionnaire
Household Food Insecurity Access Scale
Household Hunger Scale
human immunodeficiency virus
primary sampling unit
World Health Organization
Estudo do Estado Nutricional e da Dieta em Raparigas Adolescentes na Zambézia
The Academy of Finland, Embassy of Finland in Maputo, The Finnish Graduate School on Applied Bioscience: Bioengineering, Food & Nutrition, Environment, and the Finnish Cultural Foundation supported the work. We acknowledge the collaboration of Marcela Libombo in the initiation of the research collaboration between the University of Helsinki and the Mozambican partners. We also acknowledge Professor Seppo Laaksonen for help in planning the sampling design, Dr Sangita Kulathinal for useful discussions on statistical analysis and for comments on the manuscript, Elina Vaara for statistical advice, Helena Hauta-alus for her contribution to the project during and after the fieldwork, and Dr Maijaliisa Erkkola for comments on the manuscript. We are deeply grateful for all those who helped to make this study possible, especially the field workers and study participants.
None declared.