Semba RD, de Pee S, Panagides D, Poly O, Bloem MW. Risk Factors for Xerophthalmia Among Mothers and Their Children andfor Mother-Child Pairs With Xerophthalmia in Cambodia. Arch Ophthalmol. 2004;122(4):517–523. doi:10.1001/archopht.122.4.517
To characterize the risk of xerophthalmia among nonpregnant women andtheir children and the risk factors for households in which both mother andchild have xerophthalmia.
In case-control analyses of more than 15 000 households in theNational Micronutrient Survey of Cambodia, univariate and multivariate logisticregression were used to estimate odds ratios (ORs) for nonpregnant mothers,children, and mother-child pairs with xerophthalmia.
Main Outcome Measures
Risk factors for xerophthalmia.
Of 10 942 children aged 18 to 60 months and 9587 nonpregnant women,the adjusted prevalence of xerophthalmia was 0.7% and 1.9%, respectively.In multivariate analyses, a child was at higher risk of xerophthalmia whenthe mother had xerophthalmia (OR = 4.36; 95% confidence interval [CI], 2.25-8.46),and a mother was at higher risk of xerophthalmia when a child had the disease(OR = 9.21; 95% CI, 3.56-23.82). Households were at higher risk for havingboth mother and child with xerophthalmia if there was a history of diarrheain the mother (OR = 6.48; 95% CI, 1.49-28.23) or in a child younger than 60months (OR = 10.16; 95% CI, 1.55-66.62) in the last 2 weeks.
Xerophthalmia clusters among mothers and children in Cambodia and isassociated with diarrheal disease. Interventions are needed to address vitaminA deficiency and diarrheal disease at the household level.
Vitamin A deficiency is a leading cause of infectious disease morbidityand mortality in developing countries worldwide. Xerophthalmia, which includesnight blindness, Bitot spots, corneal xerophthalmia, and keratomalacia, remainsthe leading cause of blindness among children in developing countries.1 Of an estimated 453 000 children with blindnessor severe visual impairment in low-income countries worldwide, 200 000children have corneal scarring attributed mostly to vitamin A deficiency andmeasles.2 Vitamin A deficiency may occur becauseof an inadequate intake of vitamin A, an increased demand for vitamin A duringperiods of rapid growth in children, and abnormal urinary losses of vitaminA during infection. Vitamin A deficiency can result in impaired immunity,increased infectious disease morbidity, and a higher risk of corneal blindness,especially in children with measles.3,4 Xerophthalmiahas been associated with diarrheal disease5,6 andtuberculosis7 in preschool children and withgastrointestinal and genitourinary infections among pregnant women.8- 10
Recently, xerophthalmia was found to be relatively common among nonpregnantwomen in Bangladesh,11 Nepal,12 andCambodia,13 suggesting that in some parts ofthe world, women of childbearing age also constitute a group at high riskfor vitamin A deficiency. The risk of xerophthalmia among mothers and childrenin the same households has not been well characterized, and the risk factorsfor household clusters of mothers and children with xerophthalmia need furtherelucidation. We hypothesized that a mother is at higher risk for xerophthalmiawhen she has a child with xerophthalmia and that a child is at higher riskof xerophthalmia if his or her mother has this disease. We also hypothesizedthat a household cluster of xerophthalmia among mother and child was associatedwith risk factors such as low socioeconomic status and diarrheal disease.To address these hypotheses, we conducted both a cross-sectional study andcase-control analysis among nonpregnant women and their children younger than60 months in the National Micronutrient Survey of Cambodia13 in2000, a population-based survey of more than 15 000 households in 10provinces of Cambodia.
The National Micronutrient Survey of Cambodia, a collaborative effortof the Cambodian government and Helen Keller International, Phnom Penh, Cambodia,was conducted from February 2000 to September 2000. The specific aims of thesurvey were to determine the national prevalence of micronutrient deficienciessuch as vitamin A deficiency among mothers and their children younger than60 months, to assess potential risk factors for micronutrient deficiencies,and to measure current coverage of vitamin A capsule distribution programs.Data were collected from 15 120 households, which were selected by multistagecluster sampling. Cambodia's 20 rural provinces were divided into 6 groupsbased on ecological, geographical, socioeconomic, and health characteristics.From each of these 6 groups, 84 communes were chosen by randomly selectinghalf of the provinces in the group and then selecting the communes using probabilityproportional to size-sampling techniques. From each selected commune, 30 householdswith at least 1 child younger than 60 months were selected by interval sampling.Thus, the total number of households selected was 15 120 (6 groups ×84 clusters × 30 households). A household was defined as a group ofindividuals who eat from the same pot of food and, for adults in the group,who are present in the household for at least 6 months out of the year.
A standardized questionnaire in Khmer was used to obtain informationon different nutritional outcomes such as vitamin A deficiency and stunting,food consumption and vitamin A intake, demographics and socioeconomic status,and health program performance. Interviews were conducted by 40 2-person teamshired from the Cambodian government. As part of the interview, the field workersnoted the main materials of the walls of the house (bamboo, thatch, grass,hay, leaves, or other temporary materials vs wood/plywood, concrete, brick,stone, galvanized iron/aluminum, other metal sheets, asbestos cement sheets,or other permanent materials) and inquired about the amount of land that eachhousehold owned, which was then converted to hectares. The heights and weightsof women were measured using a microtoise and scale (A&D Precision HealthScale, UC-300; A&D, Tokyo, Japan), and body mass index (Quetelet) wascalculated as weight in kilograms divided by height in meters squared. Midupper arm circumference was measured with a measuring tape, and a value of230 mm was chosen as a cutoff point for undernutrition in adult women.14 Pregnancy status was determined by asking a womanwhether she was pregnant; if she was unsure, the result was coded as not pregnant.
Women were asked whether they currently had night blindness, or kwak moin ("chicken blindness"), and a parent or guardianwas asked whether each child in the household had night blindness. A historyof night blindness was shown to be a reliable indicator of vitamin A deficiencyin a study by Sommer et al,15 and throughoutour article, xerophthalmia refers to night blindness. Other questions pertainedto a history of diarrhea in the last 2 weeks and consumption of vitamin A–richfoods. Vitamin A intake was assessed with the 24-VASQ16 method,in which a 24-hour recall questionnaire was administered that included allfoods and drinks consumed during the previous day. All vitamin A–containingingredients were then assigned a food code and vitamin A content code. Thefood codes defined whether the ingredient was a vegetable, fruit, animal food,or fortified food. Vitamin A content codes were assigned for the amount ofvitamin A in the individual ingredient consumed, which was classified as lessthan 20, 20 to 75, 76 to 150, 151 to 300, 301 to 750, and more than 750 retinolequivalents. Vitamin A intake was calculated per food code using the midpointsof the vitamin A content categories. Vitamin A content of food items was obtainedfrom food composition tables available in the region (Thailand and Indonesia)that used high-performance liquid chromatography analysis for assessing carotenecontent. Vitamin A intake from fruits, including pumpkin and sweet potato,was corrected by 50% and from vegetables, including carrots, by 23% accordingto the most recent information on the bioavailability of provitamin A carotenoids.17 Thus, using the 24-VASQ method, we obtained a semiquantitativeestimate of the amount of vitamin A consumed by an individual from vegetables,fruits, or animal foods (no vitamin A–fortified foods were availablein the provinces surveyed).
For this analysis, mothers were included if they were not pregnant,had at least 1 child aged 18 to 60 months, and had a valid data entry fornight blindness for both mother and child. Children were included if theywere aged 18 to 60 months and had a valid data entry for night blindness forchild as well as mother. A case-control design was used to estimate the oddsratios (ORs) for xerophthalmia among mothers and children and then for householdclusters of xerophthalmia. All cases were selected for the analysis. Controlswere selected by calculating the number of cases for each province and usinga random sample of controls to cases in a 5:1 ratio for each province. Theratio of 5:1 was chosen to balance the number of cases and controls and tohave a large number of subjects for the analysis. Random selection was conductedusing the specific function in SPSS version 7.5 for Windows (SPSS Inc, Chicago,Ill). All analyses were conducted for the total group of cases and controls.Using a case-control design with similar methods as described previously,household clusters with xerophthalmia (cases) were also compared with householdclusters that did not have xerophthalmia (controls). Household clusters withxerophthalmia were defined as households in which both the mother and at least1 child had xerophthalmia. Household clusters without xerophthalmia were definedas households in which neither the mother nor any children in the same householdhad xerophthalmia.
Cases and controls were first compared using the χ2 testfor categorical variables. The multivariate logistic regression model includedrisk factors for which P <.10 in univariate analyses.Regression coefficients were converted to ORs and confidence intervals (CIs)for the ORs that were derived from the standard error estimates of the regressioncoefficients. The overall prevalence estimates were weighted to account fordifferences in population size of the various provinces, and the 95% CIs ofthe prevalence estimates were corrected for design effect using Epi Info version6.01 statistical software (Centers for Disease Control and Prevention, Atlanta,Ga).
There were 10 942 children and 9587 mothers in the analysis. Ofthe 10 942 children aged 18 to 60 months, 94 children had xerophthalmia,of whom 17 had a mother with xerophthalmia. Of the 9587 mothers who had atleast 1 child aged 18 to 60 months in the same household, 238 mothers hadxerophthalmia, of whom 17 had a child with xerophthalmia. There were only2 households in the entire survey in which 2 siblings aged 18 to 60 monthshad xerophthalmia. The prevalence of xerophthalmia among children aged 18to 60 months with a mother living in the same household is shown by provincein Table 1. The prevalence ofxerophthalmia was different between provinces (Pearson χ2 test; P <.001). The population-weighted prevalence of xerophthalmiawas 0.7% among children aged 18 to 60 months with a mother in the same household.The crude OR for a child to have xerophthalmia if his or her mother had xerophthalmiawas 7.2 (95% CI, 3.4-15.2).
Risk factors that appeared to be associated with xerophthalmia in childrenin the univariate analyses included a mother who had xerophthalmia duringher last pregnancy, a mother with diarrhea in the last 2 weeks, a historyof xerophthalmia in the mother, a mother with a mid upper arm circumferenceless than 230 mm, mother's parity greater than 3, a history of diarrhea inthe child in the last 2 weeks, the child not currently breastfeeding, ageof the child older than 24 months, the child not receiving a vitamin A capsulewithin the last 6 months, the child not consuming vitamin A from animal foodsduring the previous day, more than 5 family members in the household, anda roof made of thatch material (Table 2). These variables were entered into a final multivariate model,presented in Table 3. When adjustedfor other risk factors, the OR for a child to have xerophthalmia if his orher mother currently had xerophthalmia and/or had xerophthalmia during theprevious pregnancy was 4.36 (95% CI, 2.25-8.46).
Variables that were not associated with xerophthalmia among children,mothers, or mother-child pairs (P >.10) and not includedin Table 2 were the mother receivingvitamin A capsules after her last delivery, mother's body mass index lessthan 18.5, child wasted (weight-for-height z score<−2), child underweight (weight-for-age z score<−2), child's vitamin A intake from plant foods, ownership of poultryby the household, household members using a latrine, participation of anyhousehold member in a nongovernmental organization project, land owned bythe household less than 0.5 hectares, village accessible by road, and villageaccessible by water.
The prevalence of xerophthalmia among 9587 mothers who had at least1 child aged 18 to 60 months in the same household is shown by province in Table 1. The population-weighted prevalenceof xerophthalmia among these mothers was 1.9%. The crude OR for a mother tohave xerophthalmia if her child had xerophthalmia was 9.08 (95% CI, 4.10-20.0).In univariate analyses, risk factors that appeared to be associated with xerophthalmiain the mother were maternal history of xerophthalmia in the previous pregnancy,maternal history of diarrhea in the last 2 weeks, the mother not having receiveda formal education, mother's lower intake of vitamin A from plant or animalfoods, a child with diarrhea in the last 2 weeks, a child who currently hadxerophthalmia, a roof made of thatch, a wall made of bamboo, and the presenceof a home garden (Table 2). Thesevariables were entered into a final multivariate model, presented in Table 4. When adjusted for other risk factors,the OR for a mother to have xerophthalmia if her child had xerophthalmia was9.21 (95% CI, 3.56-23.82).
There were 17 household clusters in which both a mother and child hadxerophthalmia. The population-adjusted prevalence of households that containeda mother-child pair with xerophthalmia was 0.1% (Table 1). Risk factors that appeared to be associated with householdsin which a mother-child pair had xerophthalmia included a mother with a historyof xerophthalmia in the previous pregnancy, a maternal history of diarrhealdisease in the last 2 weeks, mother's parity greater than 3, a child withdiarrhea in the last 2 weeks, a child without stunting, the presence of ahome garden, and the road to the village being passable throughout the year(Table 2). These variables werefit into the multivariate model indicated in Table 5. In the final multivariate model, the risk of both motherand child having xerophthalmia was increased if the child or mother had haddiarrhea within the last 2 weeks.
This study from Cambodia shows that the risk of xerophthalmia is muchhigher for a child if the mother has xerophthalmia and for a mother if a childhas the disease. To our knowledge, this is the first study to examine therisk of xerophthalmia in mothers and children from the same households. Previousstudies have shown that the risk of xerophthalmia is higher for a child ina household where another child has xerophthalmia.18 Inthe survey in Cambodia, there were only 2 households in which siblings werereported to have xerophthalmia. Of the children who were reported to havexerophthalmia, 18% came from a household in which the mother also reporteda history of xerophthalmia. A case-control design was used within this largersurvey, as has been used in previous epidemiologic studies of risk factorsfor xerophthalmia.19,20 The prevalenceof xerophthalmia was significantly different between provinces, and we controlledfor the contributions of controls from different provinces by using this case-controldesign.
Currently the World Health Organization, UNICEF [United Nations Children'sFund], and the International Vitamin A Consultative Group have establishedcriteria for the identification of vitamin A deficiency as a public healthproblem based on the prevalence of xerophthalmia of higher than 1% among preschoolchildren21 and 5% or higher among pregnantwomen.10,22 A prevalence cutoffhas not yet been established to define vitamin A deficiency as a public healthproblem for nonpregnant women of childbearing age, as found among 1.9% ofthese mothers in Cambodia. Xerophthalmia has also been reported among 1% to2% of nonpregnant women in Bangladesh11 and6.2% of nonpregnant, lactating women in lowland Nepal.12 Thesestudies suggest that guidelines need to be developed to define vitamin A deficiencyas a public health problem among nonpregnant women in developing countries.
In our study, diarrheal disease appeared to be strongly associated withxerophthalmia in both mothers and children. This observation is consistentwith previous studies showing an association between vitamin A deficiencyand diarrheal disease.5,6,20 Longitudinalstudies in Indonesia have shown that preschool children with xerophthalmiaare at higher risk of a subsequent episode of diarrheal disease23 andthat preschool children with diarrheal disease are at higher risk of developingxerophthalmia.24 Although the relationshipsbetween diarrheal disease and xerophthalmia have not been described in a longitudinalmanner among nonpregnant women, it seems reasonable that the same epidemiologicrelationships exist among nonpregnant women as for preschool children. Theunderlying mechanisms may include a compromise in both gut immunity and gutpermeability associated with vitamin A deficiency.3,25 Ahistory of diarrhea in the mother was a significant risk factor for xerophthalmiain children, and a history of diarrhea in children was a significant riskfactor for xerophthalmia in the mother. These observations suggest that therisk of xerophthalmia increases with diarrhea among mother or child in thehousehold.
Risk factors that were associated with xerophthalmia among childrenaged 18 to 60 months also included a thatched roof, an indicator of low socioeconomicstatus. In Indonesia, xerophthalmia in preschool children was associated withlow socioeconomic status and poor hygiene.19 Nonreceiptof a vitamin A capsule within the last 6 months was also associated with anincreased risk of xerophthalmia in preschool children, and this finding reinforcesthe importance of vitamin A capsule distribution programs for children. Motherswere at higher risk for xerophthalmia if they had had the disease during aprevious pregnancy. The requirement for vitamin A increases during pregnancy,26 and vitamin A deficiency has long been recognizedas a problem among pregnant women.27,28 Thehigher risk of xerophthalmia among nonpregnant women in Cambodia suggeststhat many of these women have not been able to recover their vitamin A storesfollowing pregnancy. These findings suggest that a broader life cycle approachmust be used to ensure that women of childbearing age receive sufficient vitaminA, whether pregnant, lactating, or nonpregnant and not lactating.
Vitamin A capsule distribution programs are the main strategy for reducingvitamin A deficiency in many developing countries.1 Thesecommunity-based vertical programs are aimed primarily at preschool childrenand will miss nonpregnant women of childbearing age, another major risk groupfor xerophthalmia. Postpartum vitamin A capsule distribution to women is 1measure that may help to reduce the risk of xerophthalmia among women followingdelivery, but these programs often have limited coverage, and the World HealthOrganization–recommended dose of one 200 000-IU capsule21 to an adult is not likely to replenish maternal vitaminA stores beyond a few months. Complementary strategies such as homestead foodproduction are needed to increase the consumption of vitamin A–richfoods on the household level.29 The bioavailabilityof provitamin A carotenoids from fruits and vegetables is much lower thanhas previously been assumed,17,30 andhomestead food production should therefore focus on the development of bothanimal and plant sources of vitamin A and on generating income to improvethe dietary quality, including vitamin A content, of family members in thehousehold.31
Corresponding author and reprints: Richard D. Semba, MD, MPH, 550N Broadway, Suite 700, Baltimore, MD 21205 (e-mail: email@example.com).
Submitted for publication June 16, 2003; final revision received November5, 2003; accepted January 15, 2004.
This study was supported in part by grant HRN-A-00-98-00013-00 fromthe US Agency for International Development, Washington, DC, and grants HD30042and HD32247 from the National Institutes of Health, Bethesda, Md.
We thank Mum Bunheng, MD, and Eng Huot, MD, Ministry of Health, andHou Taing Eng, PhD, Ministry of Planning, for their support in conductingthe National Micronutrient Survey. The Ministry of Planning, Ministry of Health,and Ministry of Rural Development, Phnom Penh, Cambodia, kindly allowed theirstaffs to assist with data collection as survey team members. We thank theprovincial, district, commune, and village authorities for their logistic,management, and supervisory support, the survey teams of more than 130 personswho worked diligently to collect data throughout the country, and the membersof the more than 15 000 households for their participation and time spentwith the survey teams. We also thank the data entry department at Helen KellerInternational, Jakarta, Indonesia, for their assistance with designing, training,implementing, and supervision of data entry; the data entry operators; andMayang Sari, Msc, for design effect analyses.