Better health data with a portable microcomputer at the periphery: an anthropometric survey in Cape Verde

Bulletin of the World Health Organization, 65 (5): 651-657 (1987) © World Health Organization 1987 Better health data with a portable microcomputer at the periphery: an anthropometric survey in Cape Verde P. REITMAIER,1 A. DUPRET,2 & W. A. M. CUTTING3 A portable microcomputer wasprogrammed to record anthropometric nutritional data from children aged under 7 years in either a clinic or a population survey situation. An alarm sounds when the anthropometric measurements ofa child are below a predetermined value; an immediate check thus avoids the need for subsequent data cleaning and dis- carding of doubtful information. Data cut-offpoints in the computer can be adjusted to suit the survey or service needs of the situation. A print-out at the end of the clinic session provides immediate feedbackfor the staffand a record of the nutritional status of the group. In Cape Verde, 14 670 children attending clinics were surveyed; 26% were identified as stunted, 3% as wasted, and I % as stunted and wasted. While a portable microcomputer can improve precision, speed and motivation, nutrition surveys still depend on accurate scales, careful measurements and competent staff. Planned improvements in health care services depend on surveillance, i.e., the regular and sys- tematic collection of health-related information. The routine organization of such data is very necessary for their practical application. As long ago as 1839 William Farr, the founder of health statistics, con- sidered vital data and observed that "in order that these materials should be available for such beneficial purposes, it was obvious that they should be duly arranged" (2). Today the microcomputer can efficiently perform the necessary calculations and tabulations to present these data. Problems occur in health surveillance and data collection, even in countries with a structured health system and legal requirements to report births, deaths and notifiable diseases. In some developing countries, surveys of disease and health services are difficult to Assistant Professor, Institute of Tropical Hygiene and Public Health, University of Heidelberg, Heidelberg, Federal Republic of Germany. 2 Director, Maternal and Child Health Services, Ministry of Health, Labour and Social Affairs, Praia, Cape Verde. 3Senior Lecturer, Department of Child Life and Health, University of Edinburgh, Edinburgh EH9 IUW, Scotland. Requests for reprints should be sent to this author. carry out, but the results are essential for the effective use of limited resources. DATA COLLECTION AT THE PERIPHERY OF HEALTH SERVICES Some countries expect their understaffed health services to provide regular reports on the utilization of facilities, e.g., numbers of patients, diagnoses, deaths, and immunizations given. Often data col- lection is incomplete or inaccurate because it is a perfunctory and monotonous chore which is rarely acknowledged and the results are often not utilized when decisions have to be made about patient management or service planning. Anthropometric data on children may contain errors due to inaccurate age estimates or mistakes in measurement and recording. If calculations are involved, as when working out weight-for-height ratios, further errors are possible. "Cleaning up" such data is laborious and time-consuming and inevi- tably results in the rejection of uncorrectable results. Sometimes there is a delay of years between a survey and the publication of the results so that it is point- 4826 651 P. REITMAIER ET AL. less to base operational decisions on them. The use of a portable microcomputer is promising because it eliminates these drawbacks. EQUIPMENT AND METHOD The aim of the new approach is to collect more accurately, and collate more promptly, health data about populations during surveys, and ultimately in the health services. As an example ofhow the method and equipment can be used, the anthropometric measurements of children in Cape Verde, off West Africa, were collected at maternal and child health (MCH) clinics. A pocket computer, model Sharp PC1500A, was specially programmed to collect, calculate and store information about the ages, weights, heights, and the breast-feeding and immuni- zation status of all children attending the MCH services. The computers are battery operated and weigh only 400 g. The function of the computer is increased by a RAM-extension (Sharp CE161) and a four-colour plotter (CE150). A simple cassette recorder (CE152) serves for external mass storage. The whole outfit weighs only 2 kg making it readily transportable. One computer (used by P.R.) worked effectively for 3 years, including carriage by motorcycle and back- pack without any failures. The four-colour plotter formats diagrams, tables and lists which can be printed at the end of a clinic or working session and shared immediately with the local staff and the Fig. 1. The Sharp PC1600 pocket computer with printer. 652 USE OF PORTABLE MICROCOMPUTERS IN ANTHROPOMETRIC SURVEYS 653 families concerned. The latest model of the computer (PC 1600) utilizes a single unit for storage on disc drive and printer (Fig. 1.). The software for the nutritional surveys has been prepared for both computers, each with two separate programmes, one to process data from child clinic or outpatient services, the other from properly selected samples from any community.a The results presented here are from clinics for children. A separate paper will give details about the computing and program- ming aspects of the methodology. The following section describes how anthropometric measurements were processed and used in the Cape Verde clinics. On arrival at the clinic, the child's sex and date of birth (only those below 84 months old) are first keyed into the computer. As soon as the child has been weighed and measured, the weight and height are also entered into the computer. The anthropometric status of the child is indicated on the display. If the status shown looks unlikely to the worker doing the survey, the age, weight and height can be immedi- ately checked to see if the data entered were correct. If any of the indices (weight-for-age, weight-for- height, or height-for-age) for the individual child is outside a predetermined level, an audible alarm sounds.b This is also a signal to check again the age, weight, and height to see whether any of the data were incorrect. If these are confirmed and the child's growth status is still outside the normal range, the confirmed information is then stored in the memory. The alarm signal is also a reminder that a child whose measurements are outside the normal range deserves preliminary investigation with at least a history and physical examination to try and detect the cause ofthe poor satisfactory growth. Counselling for the parents and family will usually follow, and the child's name can be written in a register of those attending the clinic who are at high risk and need to be followed up (1). Other information, e.g., the number of immuniz- ations a child has had, can be keyed into the com- puter. If a further dose or other immunizations are indicated at this age and date, the computer's alarm could be made to sound again and the requirements will be displayed. This method is not proposed for routine service data collection at this stage, but to improve the accuracy in periodic special surveys. a Software for these programmes is available from one of the authors, Dr. P. Reitmaier. bThe predetermined cut-off points which were used for the "plausibility check" in Cape Verde were Weight-Age outside 3 standard deviations (SD), Weight-Height outside 2SD, and Height- Age outside 4SD from the reference mean, i.e., the US National Center for Health Statistics growth charts (6). The system is flexible so that the alarm can be set off at any level to suit the circumstances and availability of services for intervention. If the cut-off is set too high, so many children will be included that the services may be unable to cope. If it is set too low, children who really need inter- vention and care will not be identified. Table 1. Nutrition classification of groups of children (Waterlow/WHO) Stunted i "Normal" nutrition . -2SD + Stunted and wasted Wasted0 B t < - 2 SD Height-for-age _ RESULTS The classification of malnutrition used for the com- puter program is a slight extension of the Waterlow/ WHO method (7). The classification of children uses weight-for-height (to measure wasting) and height- for-age (to measure stunting) (Table 1.). This strict definition, using -2SD of weight-for-height as the cut-off point, classifies few children as wasted, but a considerable number ofthose who are stunted are also thin on clinical examination. In the operational situation these thin and stunted children need to be identified for appropriate intervention, and so an additional group was included in the computer pro- gram for the classification of Cape Verde children. Children who were between -1SD and -2SD below the mean for weight-for-height (slightly wasted), and were also less than -2SD for height-for-age (stunted) were described as "mildly wasted and stunted" to distinguish them from those in the classical categories of malnutrition (Table 2). In the Cape Verde clinics Table 2. Extended nutrition classification for groups of childrena Stunted (17%)I --1SD - - "Normal" nutrition Mildly wasted and stunted ((8%)1 0 Wasted and stunted (1 %) Wasted (3%) .2' -2 SD Height-for-age _'- Percentages in parentheses in each category indicate the results of the study of 14 670 children aged 1-6 years in clinics in Cape Verde. P. REITMAIER ET AL. this classification has practical usefulness, by increas- ing the proportion of children for special surveillance from about 3 % to 11% . One advantage of the com- puter is that it can be programmed to provide a cut-off point at any level, a flexibility that can be adjusted to practical action in different circumstances (see foot- note b). The proportion of children whose measurements are below a certain threshold can be printed out. Those with a weight-for-age less than -2SD are considered as underweight. This is different from the weight-for-age classification originally used by Gomez et al. (3). Using the means and SDs is a more satisfactory way to look at the nutritional status of children in a community than percentages below arbitrary cut-off points. The results for groups of children are calculated and printed out as either tables (Fig. 2, top) or diagrams indicating the means by age group, with their significance (Fig. 2, bottom). The percentages below the cut-off points are also summarized in Table 2. In Fig. 2 (top), the sex and age groups by years are shown separately. In each box relating to a nutritional parameter, the mean and standard deviation of the frequency distribution ofthe SD score for the group is at the top. Below, in small figures is the number of children in the group (N). At the bottom of the box is the percentage of the children under -2SD of the mean, or another threshold in the SD score, for the age group. At the foot of each age column is a box which contains four small figures which are a sum- mary of the types and percentages of malnourished children of that age. The bottom left figure is the percentage who are both stunted and wasted; above it the percentage who are simply wasted. The top right figure is the percentage who are "mildly wasted and stunted", and below it is the sum (rounded up or down) of all the three percentages, which gives an indication of the proportion of children in the age group who could benefit from some investigation and possible intervention. A summary of the means of the various nutritional parameters by age and sex is shown as a series of vertical bars (Fig. 2, bottom). The parallel lines at the centre of each bar indicate the mean for the group, and can be compared with the +1SD, zero, -1SD and -2SD horizontal lines. The thickness of the parallel lines is an indication of the significance, by t- test, between the mean for the group and the standard. Thin parallel lines indicate no statistical significance, while thick lines imply P<0.01 and those of inter- mediate thickness indicate P< 0.05. The data on 14 670 children attending the Cape Verde MCH clinics are shown in the tables compiled by the computer and then printed out (Fig. 2, top). Only the totals are shown here, and the nutritional state of the children is not discussed in detail since it is the data collection and compilation technique which is the main issue. The height or length of children under 12 months of age was not measured in the clinics, and therefore only the weight-for-age index is shown for the two halves of the first year. In the first six months the mean weight-for-age of the infants was positive, but by the second six months, the mean was almost -1SD, and became worse in subsequent years. The standard weight-for-height, the indicator of wasting, was depressed (but not severely) from 1 to 6 years of age; the height-for-age mean was con- sistently depressed at about -1.4SD, showing that there was considerable stunting at all ages. There appears to be more stunting and wasting in boys than in girls over 4 years of age. A total of about 11% of the Cape Verde children under 7 years of age had a nutritional status which justified health intervention. DISCUSSION Computers are gradually being introduced into decision-making in various aspects of medical care (5), and a hand-held computer has been shown to reduce unnecessary admissions to coroltary care units in the USA (4). We have made use of such an instrument to process health information about indi- viduals and groups of children in a rural Third World situation where, according to some views, there is no place for advanced technology until basic needs are met. Our experience indicates that these needs may be more effectively served by the appropriate and sensitive application of such tools. During the anthropometric survey, any measure- ment ofa child which was less than the predetermined cut-off point causes an audible alarm to sound, so that the input data could be checked immediately. Errors due to a wrong measurement or date of birth or a fault in entering the information can be reviewed and cor- rected at once while the mother and child are still present. This immediate "'plausibility check" elimin- ates the need for the stage known as "data cleaning", which is usually performed later in the process when it is impossible to verify most factors. "Cleaning" then amounts to the removal of unexplainable infor- mation and data sets that may contain the most severely abnormal children. The automatic calculation of nutritional status by the computer reduces a further source of error. The structure of a computer program standardizes the methodology and the results are more reliable be- tween different observers and in surveys at different times and places. In addition, the ease with which a computer can be commanded to use different cut-off points permits a flexibility that can be adjusted to the population needs and the service requirements. The method can be used in clinics to conduct surveys 654 0) LI) (D w r, OD C. OD 1. \.; 0! 1. T 6 Lr): 4 'H; 0 M: W (0: 4 GO:1+2 cc): 4 Lni -, ,,, + + + OD Cld(ID Ln .: .; - - (S) .: .4Ln .4 lq;j- CY) m Ln Id- M T4 T-4 r\: Ln (T): N (14: x --v M r-4: Lf-) M- ',;J- r\: N M: Ln (T); 00 4rj- x v("4: m (N: (N; m .4 W^ "-4 0 0 0 a) C4 tr) OD Lo N Cl N r\ c w N cn CD ",71r\ "If cn "7 OD. -H (.4 r\ eit ,;,r r\ in -) ") + r, + (11 (\4: + c'D LO: -v x + O-): 0) 0, (c); x x+* (N .lgu + W) C4 't W .4 m + .OD 00 .: OD .: ODcy- z 00 M (S) (D (N4 0 r\ WC r\ 11 Id-: u') m, m it (y): -v oc) it Ln: ,o- it (y): ;J- it r-\,. cc) it qd-: Ln, (y): (y), Ln:0-w - 7- T4 - (Ni M -1 z ,7A: x M I C) 0 0 0 (S) 0 0 in C4 in CD w Ln CM,09 4 CD 4 (9,4 CD r, 0) cn CDN17 40. N! IDa; m 0; C4 C4 43; IC "I-: -1 * 0-: " xto cr): T 'N U-): + Lr): x -v cq Ln: CT) :,,4 0 + N + cm m r\ 4 .4 .4 m Ln CY)Ln -:1 " ' ;j m (c toa "'-A cr): -" , U-) " m: Go " "d-i '-d- r\', x -v m F\ " 114-; (N " m: x v W 'r_4: V 4 X(N: z (N C', Z(N: X (N: V .4 ;0- r-i 0 T-4 Z-i T-4 17. 0 r-i m w cl. CD cn 0) ME cn cn I" 0 . 'f, . . ,,::U, Wt, 4'0Ln + +,(c X C!D Ln + C-4 Ln + C4 (OCID CIICY. r\+ 0 (O Ln CS) (D" C K--l 00: (O 'rr: K--l CD: M v-4; (0 13-: '-;;I- r\: N m: (c '10-: M M; I W I "-q: J It C!D 0 T-1IH CS) T-4 LO w v CD U,a) "S! CD m N,.: cn r\,R 0: + '-t-: + 4 + "q + '."4 Cs),: x + M CS) + r\, +w x xCT+ .; Ln OC) CS) 4cy- ,o Ln r\ r\ K--4 00: 'Id- it M! N p M: x-v r-A CO: M N; m ,J-: r-i CO'. '-Zi- (N: m M! C4 -1I w CS) CD 0 OD m Cn c(" OD I! ODC4 X! eD a) C(: -1 O,.qj- 4T' cnin + 0 Mi + I" (c), + + C4 ODU) 4 X X Cldm N "d-Ln .: .: 1.4 .: cn W, cY. IC-4 (s) . T-4 - (C) OC) r-4 r\,. U-) N; T) 00: x (N 0: "0- it (N: U-) (N: r-i Mi Lf-) (-4,' Ln 0:144 C" -1 N 41 w It-4 >- lr. CS) I-ri CD CD (9 + Ln: H m Ln Z (o % I o j CD 0 0 CV in dD 1, C" -4 1. v: 11cn 4N + cn U-):+ .4 CN L-N,.z Ln it Ni m (N: CS) 0 CS) CD + S E IA U) U) U) tn + + +C:). $-- i- I.- Z L!J: 0, 0: c3Z W. U) C, Z L!)', C:): IZ W, 13 L!): C3 Z 0: CIZ W:U) fA U) tn U) +' +' ixW ++ +1 ae LLi: +, + W +1 WLa W =, tn WZ LLI: Z Z Z + '. Z'W:Z -7 W: Z Z Z, + cn to <r U)" <r- D: L<,r (.r), Lc: 5:<112: Cl(f), <r-l::), <r- -7:LLI j W a W LLi ED a W LLJ <rEw- 5 E.UD: = -OD:J:X; -:Z: f.u (f): 1. -,Zi--o -LJLO I.- i-o z7w L!) LLJ Lrj LL Lo W c) LL c) W oLLJ C i -r C: <E -7- C: <1 LO 0 W LLI LLj <r- W ui LLi <r- LLi LLi LLI c 2 J a :r -Z --:Z -r u u u 655USE OF PORTABLE MICROCOMPUTERS IN ANTHROPOMETRIC SURVEYS CI cn cf l ) LL +, + 1 , < , ~~~~~~~~u I ' 6 ,|] 9 Er, I4 It, CD V- + + <1: 4~~~~~~~~~~~4 00_ ',gg E <T ~ ~~~C r~~~~~~~~~~~~~r C4 U - 0) E CL CL Z o oCD 0. o n 'D cc 0 0 E *- az (Da OC am 4- Ci ._ im en z.E_ P. REITMAIER ET AL. without disrupting the routine work. By use of standardization, realistic comparisons over a period of time and between different services are possible. The surveys in Cape Verde were conducted economi- cally and efficiently by personnel with very limited education. Accuracy of data collection can be improved by making this rather monotonous chore more interest- ing. The computer with its instantaneous response and audiovisual signals does this. The possibility of printing out the results of the measurements at the end of a session and immediately sharing them with the local population and health workers adds an import- ant dimension to the work of the data collector, since he immediately becomes a communicator of signifi- cant health-related information, not merely a clerk noting lists of numbers. This was certainly appreci- ated by the field workers in Cape Verde. If, in addition, it is possible to institute prompt action, it provides not only job satisfaction for the collector, but credibility to the process of surveillance. No amount of computer skill and programming, however, can overcome errors in anthropometric measurements. Sound staff training, checking of scales, and careful attention to detail are more important than any method of data processing. When used sensitively and intelligently with an appropriate program, a pocket computer can increase the accuracy and speed of data collection, and the motivation of staff concerned with health and nutritional surveil- lance, as was achieved in Cape Verde. ACKNOWLEDGEMENTS We are grateful for the help and encouragement of the Ministry of Health, Labour and Social Affairs, Cape Verde, the medical officers and staff including Dr I. de Carvalho and Dr E. Rocha, as well as Professor J. H. Diesfeld and colleagues at the Institute of Tropical Hygiene and Public Health, Heidelberg, Federal Republic of Germany. RESUME DE MEILLEURES DONNtES SANITAIRES GRACE A UN MICRO-ORDINATEUR PORTATIF: L'EXEMPLE D'UNE ENQUETE ANTHROPOMETRIQUE AU CAP-VERT On peut orienter les stratdgies de soins de sante primaires avec le maximum d'efficacite si l'on dispose de donnees de surveillance permettant de reperer les problemes rapide- ment et de definir les priorites en matiere d'intervention. Le micro-ordinateur est un outil precieux pour les enquetes ou la surveillance, meme dans le cadre de services de sante simplifies, car il ameliore la precision et la vitesse d'execution et renforce les motivations du personnel. Notre experience de l'utilisation d'un micro-ordinateur portatif pour enregistrer les donnees anthropometriques relatives a l'dtat nutritionnel d'enfants du Cap-Vert en est l'illus- tration. On a programme un micro-ordinateur Sharp PC1500A pour qu'il calcule, garde en memoire et classe les donnees concernant l'age, le poids et la taille d'enfants ages de 0 a 84 mois examines dans les dispensaires de sante maternelle et infantile, ou lors d'enquetes anthropometriques dans la population. Des que les donnees concernant un enfant ont ete introduites dans l'ordinateur, son etat anthropometrique est affiche et si l'un des indices est en dehors des limites preddterminees, un avertisseur sonore se declenche et l'agent de sante peut immediatement verifier les mesures et les donnees. Cet avertisseur doit etre regle de facon a indiquer les niveaux de sous-nutrition "dangereux"; en cas de declenchement, on peut proceder a un examen plus pousse de l'enfant si les circonstances le justifient et donner des conseils approprids a la famille. Les donndes peuvent etre enregistr6es sur cassette ou sur disquette. A la sortie du dispensaire ou a la fin de l'enquete, un traceur portatif fournit des tableaux et des figures illustrant l'dtat nutri- tionnel des enfants, elements qui sont immediatement montres au personnel ou aux familles, ce qui permet de discuter des problemes qui se posent et de donner des conseils pour corriger la situation. Cette approche a dte testee au Cap-Vert dans une enquete portant sur 14 670 enfants examin6s dans des dispensaires. En fixant la limite a 2 ecart-types par rapport aux normes du NCHS pour les rapports poids/taille et taille/age, on a constate 3% de cas d'emaciation, 1% de cas d'emaciation accompagnee d'un retard de croissance, et 26% de cas de retard de croissance. Au nombre des avantages de ce systeme on peut men- tionner: l'avertisseur sonore indiquant la necessite d'une etude nutritionnelle et d'une eventuelle intervention, la veri- fication immediate des mesures dans tous les cas douteux, et le fait de ne pas avoir a "epurer" les donnees par la suite pour eliminer les mesures aberrantes. Le calcul automatique des indices nutritionnels evite aussi les erreurs. Enfin, la retroaction immediate sur le personnel de sante et sur les familles montre l'importance de ces enquetes, favorise la cooperation et permet une plus grande rapidite d'action. Toutefois, les resultats dependent toujours de l'exactitude des balances, du soin apporte aux mesures, ainsi que de la competence et de la motivation du personnel. 656 USE OF PORTABLE MICROCOMPUTERS IN ANTHROPOMETRIC SURVEYS 657 REFERENCES 1. BACKETr, E. M. ET AL. The risk approach in health care with special reference to maternal and child health, including family planning. Geneva, World Health Organization, 1984 (Public Health Papers, No. 76). 2. FARR, W. First annual report ofthe Registrar-General of births, deaths and marriages in England. London, Longman, etc. for HMSO, 1839. 3. GOMEZ, F. ETAL. Mortality in second- and third-degree malnutrition. Journal oftropical paediatrics, 2: 77-81 (1956). 4. POZEN, M. W. ET AL. A predictive instrument to improve coronary-care-unit admission practices in acute ischaemic heart disease. A prospective multi- centre clinical trial. New Englandjournal ofmedicine, 310: 1273-1278 (1984). 5. SPIEGELHALTER, D. J. Computer-aided decision- making in medicine. British medical journal, 289: 567-568 (1984). 6. UNITED STATES NATIONAL CENTER FOR HEALTH STATISTICS. Growth charts. Rockville, MD, US Department of Health, Education and Welfare, Public Health Service, Health Resources Administration, 1976 (HRA 76-1200, 25, 3). 7. WATERLOW, J. C. ET AL. The presentation and use of height and weight data for comparing the nutritional status of groups of children under the age of 10 years. Bulletin of the World Health Organization, 55: 489-498 (1977).