-360-
A MODEL PROTOCOL FOR A DIABETES AND OTHER NONCOMMUNICABLE DISEASE FIELD SURVEY Gary K. Dowse• & Paul Zimmetb
Introduction Field surveys of diabetes and other noncommunicable diseases (NCD) (mainly hypertension and coronary heart disease), plus the risk factors for these diseases, are increasingly being performed in both developed and developing countries (1). Such surveys are conducted both in the context of academic etiological enquiry and for public health-related documentation and planning purposes, spurred on by the growing importance of these diseases in economic and social terms. While suggested methodologies for aspects of such surveys have been previously published {2-4), there are no readily available practical examples of model protocols for diabetes and NCD surveys which might be used as a guide by investigators planning to develop their own field survey strategies. To address this perceived need we present here a model protocol and survey manual which has been the basis for surveys performed successfully in recent years in both rural and urban areas in many developing countries of the Oceania region (5-7) and in the newly-industrialized nation of Mauritius {8-10). lt has also been modified recently for use in the developed nation of Singapore. Readers should note that this protocol should not be seen as the "gold standard": it represents the writers' cumulative experience and includes some compromises in technique necessary in the context of the particular environments for which it has been developed. Individual workers should see fit to modify it to suit their own circumstances and other advice. Where particular alternatives should be considered, a note appears in smaller type. Aims 1. General 1.1 A survey of the adult population aged ..... [e.g. 30-64 years] residing in ..... [e.g. geographically defined areas] aimed at defining the current epidemiology of non-insulin-dependent diabetes mellitus (NIDDM) and cardiovascular diseases (CVD) and their associated risk factors. 1.2 [Example for longitudinal study]. A health status followup of the individuals who participated in the ..... [e.g. 19871 Noncommunicable Disease and Risk Factor Prevalence Study.
ance (IGT), NIDDM, hypertension and coronary heart disease). 2.2 To describe the distribution and determinants of risk factors levels (e.g. obesity, cigarette smoking, serum lipids), and health knowledge, attitudes and behaviours. 2.3 To define the prevalence and associated risk factors for complications (e.g. retinopathy, microalbuminuria and proteinuria, peripheral neuropathy, and obstetric end-points such as stillbirths and congenital malformations, in subjects with diabetes and IGT). [Examples 2.4- 2 7 apply to long1tudmal follow-up studies.]
2.4 To describe the pattern of mortality and relate this to disease and risk-factor status defined at baseline. 2.5 To study the incidence and determinants of IGT, NIDDM, hypertension and coronary heart disease, and their risk factors, in the group of individuals studied in both ..... [year al and ..... [year b].
2.6 To study the natural history of glucose intolerance in the group of subjects for whom previous glucose tolerance data are available. 2.7 To compare risk-factor levels, health knowledge and attitudes, and dietary practice between the intervention and reference communities to determine the effect of health intervention strategies over the period a - b. Sampling The nature and size of a sample for a survey of diabetes and other noncommunicable diseases will vary depending on resources, aims and logistic aspects such as geography and transportation systems. While a simple random sample of a defined population might be the ultimate goal, this is rarely possible. The majority of NCD surveys are performed in selected community samples ("purposive" sampling), usually chosen because of convenience and/or because they are judged to be "typical" given the a priori objectives of the investigators (e.g. comparing rural and urban dwellers) (4,5). In other situations, random sampling of population "clusters" already defined for administrative/electoral reasons can be undertaken (4,8). Stratification, with oversampling of a particular sub-group (such as a minority ethnic group or age-group) and/or randomization (usually "systematic", whereby, for example, every nth household might be chosen) can be added as additional stages to either community samples selected purposively or randomly-selected cluster samples. However, particularly when working in developing countries, aside from being logistically simpler, response rates are likely to be higher when all subjects within a defined community/geographic area are invited to participate. More detailed considerations of sampling and statistical aspects, including calculation of sample size, can be found elsewhere (4, 11). Rapp tnmest stat1st.
2. Specific 2.1 To define the prevalence and associated risk factors for disease (e.g. impaired glucose toler'Ep1dem1ologist, International D1abetes Institute, Melbourne, Australia. b Director, International Diabetes Institute, Melbourne, Australia.
samt mond., 45 ( 1992)
-361Census, invitations and advertising [The following example applies to the situat1on where whole community clusters have been chosen, either purposively or randomly.]
Survey procedure and measurement techniques An appropriate survey site should be selected for each location. The nature of the survey site chosen will depend on factors such as proximity to survey population, availability, facilities (e.g. toilets, running water, electricity, quiet/private rooms for certain procedures), and size required in relation to range of procedures and number of subjects invited. Schools (in holiday time) are particularly suitable as subjects can progress from room to room where successive procedures are located, but community halls, health centres or even open outdoor areas may be utilized. A convenient starting time each morning could be 0800 hours (arrival 0730). In some communities, an earlier starting time may be preferred. Work for the day _shou_ld be largely completed by early afternoon, l~avmg t1me for preparation for the following day (mcluding re-invitations to non-responders and subjects selected for additional screening for com~lications, etc.). In the afternoon following completion o_f work in a particular location, equipment can be sh1fted to the next survey site and made ready. Work should proceed 5 or 6 days per week (See Table 7). A survey team may be made up of approximately 15-20 members, distributed according to task as outlined in Fig. 7. Further staff will be involved in completion of biochemistry, driving, assisting with complications screenings, and in performing data entry. However, the number of staff required will vary depending on local circumstances and upon the budget. For the core component of the model survey described here, around 17 team members could proce~s approxi.n:tately 120-140 subjects comfortably by m1dday. Cnt1cal factors which determine the number of staff required include the number of subjects invited per day, the time per procedure, and the number of procedures. Given that subjects come fasting for the glucose tolerance test one should aim for t~e las~ subject to have completed all proce_dures, mcludmg the 2-hour blood collection, by midday_. If the number of staffing available is limited, then e1ther fewer subjects should be invited, or fewer procedures should be undertaken. the above arrangement may be ideal, it is poss1ble to undertake a more limited survey successfully with a much smaller team. For example, 5-6 team. me~bers c~>Uid_ process 60-100 persons per mornmg m the s1tuat1on where fasting blood was ~ot ta~en, ECGs were deleted, only a limited quest1onn_a1re was undertaken, and staff doubled-up on certam procedures (Fig. 2). Under this scenario all subjects would need to be registered strictly with,in a 2-hour t!me-frame, so as to allow staff undertaking the earlier procedures to move on to venesection and laboratory duties once the time for 2-hour blood tests fell due. Of course, the ultimate limitation of ~taff required might see only 1 investigator performmg all procedures, but this would necessarily allow only a handful of subjects to be surveyed per day. Survey procedures should be undertaken in the following sequence (See also Fig. 7). 1. Registration (2 core staff plus the census officer for each location) (a) Greet the subject. (b) Add the subject's name to the daily register and allocate the next survey number. Alth~ugh
A house-to-house census of the defined areas should be completed before the survey begins. The names of all "usual residents" aged ..... years should be recorded. lt is usually sufficient for a reliable informant (such as the head of the household) to give details of the individuals absent at the time of the census. [Spec1fic instructions should be prepared for census officers.] At the time of the census, written and verbal information should be given to each household concerning the reasons for the survey and what it might entail for each individual. There should be a specific "motivator" for each survey area who has primary responsibility for liaison in that area and for issuing invitations and re-invitations (e.g. for initial non-responders and special follow-up investigations such as retinal photography). As well as initial contact at the time of the census, each community should be contacted again several days before the arrival of the survey team in that area. Additionally, all individuals should be given an invitation letter (with instructions for fasting) within 2-3 days of their appointment. Prominent members of the community (e.g. religious leaders, village mayors) should be recruited to help with promoting the survey. Newspaper advertisements, radio and television announcements (where appropriate) and posters in survey areas are also helpful. Subjects who require it should be given an official letter for their employer requesting the latter's cooperation in allowing the person time off from work to attend the survey (plus a certificate of attendance at such time as they do attend). Non-response The extent to which non-responders should be sought will depend on the overall response rate, but would normally involve at least one re-invitation while the survey team remains in the area. Demographic characteristics of non-responders (as determined at the household census) should be compared with those of responders. Where possible, all or at least a sub-sample of non-responders should be contacted following the study to complete a brief questionnaire (a) to identify their reasons for nonattendance (e.g. too ill, dead, not interested, away from area, etc), and (b) to assess the frequency of self-reported major end-points of interest such as diabetes or hypertension. Comparison of the latter with frequencies in the responder sample can help in the detection of obvious response bias.
Schedule, accommodation and transportation The survey should begin on ..... [datel and run for a period of ..... lxl weeks, assuming that approximately ..... lyl persons are surveyed per day with an estimated total survey population of ..... [zl persons. The census of the survey areas should be completed at least ..... [e.g 3 weeks] prior to the starting date. [If appropriate, also give details regarding any accommodation and transportation requirements for staff during the period of the survey.]
Wld hlth stattst. quart., 45 (19921
-362TABLE 1. TABLEAU 1. EXAMPLE SCHEDULE FOR NONCOMMUNICABLE DISEASES STUDY
EXEMPLE DE CALENDRIER POUR L'ETUDE DES MALADIES NON TRANSMISSIBLES April1992- avril1992
Date
Locat1on- l1eu
April- avril
6 7 8 9 10 11 12 13 14 15 16 17 18 19
Mon - lundi 6========================= OFF- ARRET================================= Tue- mardi 7 Site 1 Wed - mercredi 8 Thurs- jeudi 9 Fri- vendredi 10 Sat- samedi 11 Sun- dimanche 12 ===================== OFF - ARRET================================= Mon- lund1 13 Site 2 Tues- mardi 14 Wed- mercredi 15 Thurs- jeudi 16 Fri- vendredi 17 Site 3 Sat- samedi 18 Sun- dimanche 19 ===================== OFF- ARRET=================================
20 Mon- lundi 20 21 Tues- mardi 21
etc.
Example lay-out of daily registration book: Survey number 39 40
2. Fasting blood sample (3 core staff plus provision for a fourth during peak periods) Subjects should be seated and the specimen taken from the cubital fossa of the chosen arm by standard technique, pref~rably using vacuum tubes, on which the subject's survey number only is recorded clearly prior to venepuncture. The tourniquet should first be applied, followed by skin cleansing with an alcohol or spirit swab. On successful venous entry, the tourniquet should be released if possible and the required vacuum tubes carefully inserted and removed. A cotton swab should be applied to the site as the needle is removed, and the subject instructed to maintain pressure on this for two minutes with the arm in extension. Tape should be provided for use where necessary. Tubes should be gently inverted a few times and then placed in the racks provided at each bleeding station, ready for collection by laboratory staff. An "F" should precede the survey number (eg F138) on all fasting tubes. [As an example, 10 ml plain and 2 ml draw fluoride/oxalate vacuum tubes can be used for all subJects: 3 aliquots of serum can be taken from the 10 ml tube: one for lipids and other routine b1ochem1stry, one for insulin determination, and a third for storage. Specimens for glucose determination must be collected into fluoridated tubes. If whole blood glucose is to be measured then the fluoridated tubes should be centnfuged immediately, certainly within half an hour, and the plasma separated as it IS known that glycolysis results 1n a 5-15% fall in glucose concentration over 2 to 4 hours, even 1n the presence of fluoride. If 1mmediate centrifugation is not possible then the fluoridated tubes should be kept in a cool box on ice at 0-4°C until such t1me as they are centrifuged.]
Gtven name John Mary
Surname Sm1th Brown
Comments
and so on
Tick the subject's name off on the household census list and add his or her survey number to the appropriate column. (d) Transcribe the survey number and the household number (available from the census list) to the individual's survey sheet. (e) Complete section 1 (Demography) and section 2 (GTT and diabetes history) of the survey form (see AppendiX,. Of the two permanent registration staff one should complete the demography section and one the GTI status section. Subjects determined at this stage to be taking oral drugs or insulin currently and regularly should not have a glucose load. All other subjects (including those who report being diabetic but not currently on treatment) should have a glucose load. (f) Direct subjects to the bleeding station. (c) [Where additional studies are performed on a sub-sample of survey participants (e.g. supplementary blood collection, procedure or questionnaire) it is convenient to identify them at the registration desk by placmg, for instance, a coloured sticker on a pre-defined part of the1r survey form. This immediately alerts other staff that this subject belongs in the additional study.] 'Th1s is the offic1al recommendation of WHO (12). However, many population studies have been performed using 75 g of glucose (dextrose) monohydrate, wh1ch is more conven1ent for use 1n the field. Unfortunately, 75 g of the latter is equivalent to only approximately 68 g of anhydrous glucose, so a disparity has arisen between the recommendations and common practice. Currently it is unclear wh1ch size load is the most accepted. Investigators should state clearly wh1ch they have used in a particular study, and future studies should probably follow the WHO recommendations from the outset, for 75 g of anhydrous glucose or equivalent (approximately 82.5 g of glucose monohydrate). Follow-up of populat1ons formerly tested with 75 g of glucose monohydrate should probably opt for consistency with the original examination.
3. Glucose load ( 1 core staff) (a)
250-300 ml of solution (containing 75 g of anhydrous glucose or equivalent)c should be measured into plastic re-usable cups from a measuring cylinder. This should be drunk within 5 minutes, commencing as soon as possible after blood collection. lt should not be given to subjects who currently take oral hypoglycaemic tablets or Rapp tnmest. statist santt mond, 45 11992)
-363FIG. 1 MODEL SURVEY FLOW-SHEET FOR A LARGE-SCALE DIABETES AND CARDIOVASCULAR DISEASE FIELD SURVEY, REQUIRING 15-20 TEAM MEMBERS AND ALLOWING 120-140 SUBJECTS TO BE ASSESSED PER DAY.
A GRANDE ECHELLE SUR LE DIABETE ET LES MALADIES CARDIO-VASCULAIRES, AVEC UNE EQUIPE DE 15 A 20 PERSONNES CHARGEES D'EXAMINER DE 120 A 140 SUJETS PAR JOUR. Registration (census taker and 2team)
MODELE DE DIAGRAMME POUR UNE ENQUETE SUR LE TERRAIN
I Enregistrement (recenseur et equips de 2 personnes) I Glycemia
Fasting blood (3 team)
a jeun (equipe de 3 personnes)
Glucose load (1 team)
Charge de glucose (1 personne)
Height & weight (1 team)
I I I
Po1ds et taille {1 personne)
Waist & hip girth (1-2team)
Tour de taille et tour de hanche (1-2 personnes)
Questionnaire (2 team)
Questionnaire (2 personnes)
• • • • • ~
I I
I I -1 j
Subsample of subjects for special studies such as measurement of Jean body mass, physical activity questionnaire
., ILaboratoire (2-4 personnes) I ~
Sous-echantillon de sujets pour etudes speciales, telles que mesure du paid s maigre, questionnaire sur l'activite physique
I Blood pressure (2 team)
_j
I I
Tension arterielle (2 personnes)
ECG (2team)
Electrocardiogramme (2 personnes)
I I
2-hour blood (3 team)
Glycemia 2 heures apres le repas (3 personnes)
+ ~
Registration {check record sheet)
I
Enregistrement (verification du fichier) WH0921080
I
insulin regularly for treatment. Fresh water should be available for rinsing cups before reuse. (b)
Example lay-out of glucose load book: Survey number 39 40 Given name John Mary
Surname Smith Brown
A battery operated digital clock should be maintained at the glucose station. A sequential listing in the glucose load book of subject survey number, name, and the time for 2-hour blood collection (the clock may be set two hours ahead of local time) should be maintained. This time should also be recorded clearly in indelible ink on the back of the subject's hand (or on an adhesive label attached to the clothing), and the importance of this should be explained to the individual. The time recorded will be the time at which the subject begins to drink the glucose+ 2 hours.
2-hour time 1001 1002
Comments
and so on
(c)
Put a tick or check mark in the box on page
3 of the survey form (see Appendix) if the (d)
load is given, but a cross or "x" if not. The glucose load officer should be responsible for ensuring that subjects return for their 2-hou r collections at the appropriate time, and this will involve collaboration with the phlebotomist, who will also have a clock/watch synchronized to that at the glucose load station. As subjects have their 2-hour blood taken a tick should be placed
Wld hlth stat1st quart., 45 ( 1992)
-364FIG.2 MODEL SURVEY FLOW-SHEET FOR A MODEST DIABETES AND CARDIOVASCULAR DISEASE FIELD SURVEY UNDERTAKEN WITH MORE LIMITED RESOURCES
(A·E refer to 5 specific team members and demonstrate how a team of this size can undertake more than one task in order to examine 60-100 subjects per day.) MODELE DE DIAGRAMME POUR UNE ENQUETE SUR LE TERRAIN A MOYENNE ECHELLE SUR LE DIABETE ET LES MALADIES CARDIO-VASCULAIRES, REALISE AVEC DES RESSOURCES PLUS LIMITEES
(Les rubriques A a E font intervenir 5 personnes differentes pouvant executer plusieurs taches de maniere a examiner de 60 a 100 sujets par jour.)
Registration & census taker (A) I Enregistrement, recenseur (A)
Glucose load (B) I Charge de glucose (B)
Height & weight I Taille et poids Waist & hip girth (C) I Tour de taille et tour de hanche (C)
Questionnaire (D) I Questionnaire (D)
Blood pressure (E) 1 Tension arterielle (E)
2-hour blood (B, D) I Glycemia 2 heures apres le repas (B, D)
Laboratory (A, C) I Laboratoire (A, C) WH0921081
in the "Comments" column of the glucose load book. If subjects are late for their 2-hour collection, this fact should be noted in the "load" book. The time at which the 2-hour blood is actually taken should be recorded on the survey form by the phlebotomist. (e) An appropriate system should be developed for preparation of fresh supplies of glucose solution on the evening prior to each survey day. 4. Height and weight (1 core staff) Subjects should be in light clothing, without shoes. (b) Height should be recorded to the nearest centimetre, rounding up if midway, using a measuring rod. Subjects should stand upright with back against the stand, heels together and eyes directed forward so that the top of the tragus of the ear is horizontal with the inferior orbital margin, and the measuring plate lowered onto the scalp to give the correct level. (c) Weight should be recorded to the nearest 0.1 kilogram, rounding up if midway, using the available scales, which should be on a firm, horizontal surface. The zero should be (a)
checked each day and calibration should be crudely assessed by reference to an individual of "known" weight (this is particularly important when the scale is transported to another survey site). 5. Waist and hip circumferences (1 or 2 core staff) (a)
If necessary, there should be male and female observers for respective sexes. Subjects should stand relaxed in a screened area. One layer of light clothing over underwear is acceptable. The observer should kneel or sit at an appropriate height in front of the subject, who should breathe quietly and normally. (b) A dress-maker's measuring tape should be used, taking care that it is applied horizontally. (c) Waist girth should be measured at the midpoint between the iliac crest and the lower margin of the ribs. An approximate indicator of this level may be ascertained by asking the subject to bend sideways. (d) Hip girth should be recorded as the maximum circumference around the buttocks posteriorly and indicated anteriorly by the symphysis pubis. Rapp tnmest. stat1st samt mond, 45 (19921
(e)
365 ment recorded. A further measurement should then be taken by following the same sequence. lt is important that the mercury column is at the eye level of the observer during measurement, and that readings are taken at the top of the level of the meniscus, to the nearest 2 mmHg, rounding up if halfway. If the observer is unable to hear or forgets a phase recording, the cuff should be deflated and the measure repeated. In cases where there is no apparent phase 5, the phase 4 measurement (first muffling of sound) should be recorded, with such noted on the record sheet adjacent to the reading. If the two readings are different by greater than around 20%, then a third measure should be recorded adjacent to the others.
Measures should be made to the nearest 0.5 centimetre (rounding up if necessary) and repeated following both initial recordings. If there is variation greater than 2 cm between duplicate readings then a third should be taken and recorded alongside the second. (In these cases the 2 most consistent readings can be used in analyses.)
(f)
6. Questionnaire (2 core staff) Questions should be asked in a standard manner using the local language, and numbers recorded legibly in pencil in appropriate boxes. Each box on the survey form should have something written in it (see Appendix). A blank box indicates that the question mistakenly has not been asked. If a question is inapplicable to that subject (e.g. medications for diabetes in a non-diabetic, obstetric history in a male) then place a diagonal line through the boxes as indicated IZI. Specific instructions regarding completion of the questionnaire should be given to the staff involved. 7. Blood pressure (2 core staff plus reserve) Subjects who have rested in a seated position during and after questionnaire administration should move only a short distance to the blood pressure station. This should be a quiet area. Standard mercury sphygmomanometersd should be used. (a) Outer garments should be removed to properly expose the right arm. If the latter is missing or deformed, then the left should be used, and this noted adjacent to the blood pressure records. (b) The right arm should rest comfortably on the table, elbow level with the heart, and upper arm at an angle of about 40° to the trunk. (c) Large adult-size cuffsd should be used as standard. The cuff should be applied firmly with the middle portion of the bladder positioned over the brachial artery. The lower edge of the cuff should be 2-3 cm above the cubital fossa, to allow space for the bell of the stethoscope. (d) Establish the pulse obliteration pressure by palpating the radial pulse with the fingers of the left hand while inflating the cuff with the other. Inflate the cuff to about 30 mmHg above this level. (e) Place the stethoscope bell lightly over the position of the brachial artery. Release the cuff pressure at a steady rate of about 2 mm per heart beat. Systolic pressure is the level where the first sounds identifiable as pulses are heard. Diastolic pressure should be taken at the level where sounds cease (5th phase). The cuff should then be deflated completely and the measure-
8. ECG (2 core staff) (Recording of electrocardiograms and class1f1cation according to the Minnesota Code (2,4) is regarded as an objeCtive method of assessing the frequency of ischaemiC heart disease in populations, particularly where routine mortality data and special registers of myocardial infarction are unreliable or non-existent. However, the method IS relatively insensitive and lacks specific1ty, particularly for ST/T wave abnormalities in females, and in populations where there is complete death certification and clmical record-keeping, it may be unnecessary to mclude ECG exammat1on in surveys. Questionnaires for assessing the frequency of angina and myocardial infarction have been developed and validated for use in some populations and should also be considered as an addition to, or alternative, to the use of ECGs (4). If ECGs are recorded, multi-channel machines with battery back-up are preferred. At least five complexes should be recorded for each lead and it is important to ensure a stable base-line to facilitate coding.]
(a)
Resting 12-lead ECGs should be performed on subjects aged 35-64 years only. There should preferably be a quiet area with a screened couch. !Eff1ciency can be improved by having two couches separated by a screen, with the machine located centrally at one end. One subject can be prepared while the other's ECG is in progress.]
(b)
(c)
Careful preparation of the skin-electrode contact is necessary, involving skin cleansing and shaving, application of jelly, and use of clean electrodes with firm cable contacts. Clip-on limb leads are convenient and time-saving. Chest leads should be positioned in a standard fashion and the subject's name and survey number (preprepared sticky labels may be provided for the latter) recorded on his/her tracing. The box on the subject's record sheet should be ticked to indicate that the test has been performed. Following completion of the ECG, excess jelly should be wiped from the skin and the recordings maintained at this station in order of survey number.
9. Two-hour blood sample Collection method as for fasting specimen, with the subject number clearly preceded by a "T" (e.g. T352). (As a m1n1mum, a 2 ml draw fluoride/oxalate tube should be used, and additionally a 5 ml plain tube if serum insulin is to be measured.]
There are varymg opinions on whether random zero sphygmomanometers (2,4) are preferable to standard models in population surveys. Based on their own experience the authors have now chosen to use only standard machines, but emphasize the importance of proper training of blood pressure observers. Similarly, opinion on cuff/bladder size differ: the authors, who often examine relat1vely obese populat1ons, have chosen a smgle long cuff which ensures a full encircling of the arm in all subjects. d
This procedure, and its correct timing, is critical in a diabetes survey. Staff should collaborate to ensure that subjects are marshalled and available several minutes prior to the due time, which is recorded both in the glucose load book and on the back of the subject's hand (or on an adhesive label on their clothing). The phlebotomist should record the time at which the
Wld hlth statlst. quart., 45 (1992)
-
366 such as membranes and probes. Glucose values and subject numbers should be recorded clearly in separate columns in the Glucose Results Book and transcribed to survey forms as soon as possible at the end of the day's work. Every 10th PG should be performed in duplicate, the plasma being saved for further external quality control checks. Additional quality assurance will result from the use of solutions of known glucose concentration at regular intervals, independent of the standards used for calibration. (c) The glucose analyser technician and the assistants should ensure that: (i) glucose specimens are collected from the venesection station and centrifuged as quickly as possible (within 30 minutes). Plasma should be pipetted into microcentrifuge tubes and presented to the technician. (ii) 10 ml (fasting) and 5 ml (2-hour) plain specimens should be left at room temperature and allowed to clot prior to serum separation. (iii) First morning urine specimenS' may also be collected from subjects with diabetes or impaired glucose tolerance, plus a control group. Collection jars and instructions should be issued to the selected individuals within 1-2 days of their glucose tolerance test. Specimens must be centrifuged prior to storage. (d) At the end of the survey activities each day, specimens should be transported to the laboratory where routine biochemistry is to be performed. Other aliquots of sera (and urine), should be stored frozen at -20 (or -70 °C) prior to air transport at the conclusion of the survey. An outline of specimen collection and handling appears as Box 1.
two-hour specimen is actually taken in the "check" section of the survey form. Where a subject is 10 or more minutes late for the 2-hour test, this should be noted in the comments column of the load book. Note that diabetic subjects taking oral hypoglycaemic tablets or insulin (who should not have been given the glucose load) will not require a two-hour specimen. 10. Final check At the conclusion of a subject's procedures (this should normally be the 2-hour blood collection station) the subject should be asked to return to the registration area where the survey form can be checked for completeness, and they can be thanked for participating. Basic results (blood pressure, plasma glucose, weight) should normally be issued to individuals within a few days, with advice regarding follow-up where necessary. lt may be convenient also to issue lists of survey numbers to each examination station, to be ticked after each subject completes the examination. In this way, subjects failing to report for a particular examination will be readily identified. 11. Laboratory (2-4 core staff) [The followmg example relates specifically to the situation where plasma glucose is measured at the survey site while the survey is in progress. The authors prefer th1s arrangement because: (a) subjects can be g1ven the1r results within a day or two, facilitating therapeutic action in those w1th abnormal values; (b) sub·stud1es dependent on knowledge of glucose concentrations can proceed; and (c) there IS no nsk of loss of critical samples dunng shipment or due to freezer failure. A Yellow Springs Instrument (YSI) Model 23AM machine is the bench-mark for accurate and precise measurement of glucose in the field, and can be run from a generator or even car battenes if necessary. Unfortunately, this model IS no longer manufactured, and a suitable alternative has yet to be found although a new "portable" YSI has recently appeared on the market. Reflectance meters are in general not so reliable or accurate for population studies although the Reflotron and Reflolux (Boehringer Mannheim) have been commended by some investigators. Other new models include the Hemocal and the Analox. Regardless of which machine is selected, meticulous calibration and quality assurance is required and the importance of careful trammg and certif1cat1on of technicians cannot be over-emphas1zed. In some situations it may be necessary to freeze plasma collected in the field and measure the glucose some days or weeks later, perhaps after air-shipment with dry ice. In developed countnes and in urban areas of less developed countries there may be ready access to established laboratory facilities where, provided that plasma is separated immediately at the survey site and kept cool, fresh samples m1ght be transported Within 24 hours or so for glucose measurement using a variety of automated multi-sample and dedicated single-sample machines.] ( 13)
oc
12. Special studies [The following examples have successfully been incorporated into the authors' studies when add1t1onal staff and facilit1es were available.]
(a)
(a)
Temporary laboratory space should be established at each survey site with bench space, centrifuge(s) and glucose analyser(s). This facility should be in reasonably close proximity to the area of blood collection. (b) One person should be responsible for measurement of the plasma glucose, and the calibration (with 10 and 25 mmol/1 standards after every nth specimen) and maintenance of the glucose analyser. There should be at least one reserve machine in case of breakdown, as well as spare parts
'Timed overnight unne collections for calculation of the album m excretion rate are difficult to perform within the setting of a field survey. "Spot", or preferably first morning specimens and measurement of albumin concentration, or albumin/creatinine ratio, represent a reasonable compromise where necessary.
Retinal photography: Subjects with diabetes (known or newly-detected during the survey) and a sub-sample of subjects with IGT (e.g. 1 in 4) should be given appointment times. Transportation should be provided from the local survey site to the site for photography, and back again. Visual acuity should be assessed prior to photography, as well as in a sub-sample of the main survey population. (b) Urinary microalbumin: All subjects with diabetes, plus a sub-sample of subjects with IGT (e.g. 1 in 4) and a group of normal controls (e.g. multiples of 20 who also have normal glucose tolerance), should be given a urine collection jar (labelled by name and survey number) and a suitable paper bag within 1-2 days following their glucose tolerance test. They should be asked to fill the jar with a clean sample of their first morning urine on the following day. In most cases this should coincide with an appointment for retinal photography and the specimens will be received either at the time people are collected at home, or when they Rapp. tnmest. stattst samt. mond., 45 (1992)
-367Box 1.
Example of specimen-handling instructions for a noncommunicable disease survey Storage DestinatiOn
CollectiOn tube
Treatment
Test
Blood: fasting (a) Fluoride/oxalate 2ml draw (i) 2 500 rpm/10 min. a.s.a.p. (ii) Measure glucose on-site 2 500 rpm/10 min. after standing
1/10-7- 1ml@ -20°C (i) 2 ml serum cooi/-20°C 1/10-? 1 ml @ -20°C (ii) 1.5 ml -20°C/-70°C (iv) 1.5 ml -20°C (i) - 5 ml plasma @ -70°C (ii) Buffy coats @ -70°C 2.5 ml plasma @ -70°C
Newcastle Central Lab. Newcastle Newcastle Melbourne Canberra St. Louis Newcastle
Glucose QA Lipids, etc. Lipids QA Insulin Storage Genetic markers Candidate gene studies PA/1, other haemostatic factors
(b) 10 ml plain
(c) 10 ml heparin (sub-sample)
2 500 rpm/10 min. Keep cool/separate
(d) 5 ml sodium citrate (sub-sample)
2 500 rpm/10 min. Keep cool/separate
Blood: 2 hours after oral glucose (e) Fluoride/oxalate 2 ml draw 1) 2 500 rpm/10 min. a.s.a.p. 2) Measure glucose on-site 2 500 rpm/10 min. after standing
1/10-? - 1 ml @' -20°C 2.5 ml @ -20°C
Newcastle Newcastle
Glucose QA Insulin
(f) 5 ml plain
Urine (g) 70 ml jar (sub-sample) 2 500 rpm/10 min. 5 ml @' -20°C Melbourne Microalbumin/ Creatinine ratio
assemble at the survey site to await transportation to the place of photography. Specimens should be kept cool until such time that they can be centrifuged and a 5 ml aliquot is stored and frozen. (c) Vibration threshold: The same group of subjects attending for retinal photography (see above) may have vibration thresholds assessed using a commercially available electrical instrument. A group of controls with normal glucose tolerance should also be tested. (d) Lean body mass: This should be determined at each survey site on subjects aged 35 years and above using a commercial available spectrophotometer. (e) Diet/health knowledge and attitudes: These may be determined on a computergenerated age-sex stratified sub-sample of the survey population by trained interviewers, following the conclusion of the main survey. (f) Genetic factors: Options for genetic research within the type of field survey described here are limited. In the past, associations have been sought between disease (e.g. NIDDM) and a wide range of serological genetic markers readily measured in blood collected during surveys. If additional resources are available, investigators may wish to pursue family studies, if suitable families are identified during the survey. More recently, buffy coats have been collected and stored during field surveys, ready for later harvesting of DNA and study of Wld hlth statist quart, 45 ( 1992)
associations between disease and specific genes. In this context, a reasonable approach in a diabetes and other noncommunicable disease field survey of NIDDM is to collect an additional blood specimen specifically for a genetic case-control study in subjects with known disease, and in older subjects (e.g. > 50 years) with no history of disease, at the time of the fasting blood collection. If some of the latter group are found to have NIDDM in the glucose tolerance test, they can be re-allocated to the "case" group. Subjects to have the additional blood collection can be identified to the phlebotomist by placement of a coloured sticker on a prominent part of the survey form by registration staff.
13. Quality control Stringent attempts must be made to ensure the validity of data, and this implies rigorous standardization of methods and both internal and external quality checks. Observer variation can be minimized by limiting the number of observers performing each task in the survey, and by ensuring that there are no systematic differences in the manner in which groups of subjects present to particular observers. The following steps should be taken: (a)
Initial training of observers for each procedure. In the week before commencement of the survey, measurement techniques will be refined on groups of volunteers (e.g. for
-
368 (c)
anthropometry and blood pressure) and repeatability assessed. [Wherever possible, blood pressure observers should be formally trained using video or audio-tapes and a double-headed stethoscope with a certified "gold-standard" observer. Training systems are referenced in Rose et al. (4)] (b) Regular checking and calibration (where appropriate) of measuring equipment. (c) Biochemistry: internal and external quality assurance should be built in, as described earlier for glucose.
Range and consistency checks should be undertaken on all data, which should be double-entered before production of the final verified and cleaned data-set.
16. Data analysis Current prevalence of disease and risk factors e.g. NIDDM (known and/or newly diagnosed), IGT, cigarette smoking, obesity, hypertension, and ECG ischaemia, by age, sex and ethnicity should be calculated. Standardization (e.g. for age) should be performed where appropriate. Incidence and trends in prevalence should be determined using longitudinal data. Mean/median values and distributions of continuously distributed variables should be described. (b) Simple statistics (e.g. X2 and t-tests) can be used to identify differences between groups. (c) Relative risk estimates for risk factordisease associations can be determined using Mantei-Haenszel techniques. (d) Multivariate statistics, including multiple linear and logistic regression can be used in attempts to elaborate important independent risk factors in cross-sectional and longitudinal data. (a)
14. Ethics [A statement should be included detailing the authont1es who have given ethical approval for the study to proceed. Researchers working in foreign countries should normally have approval from both their country of ongm and that in which they will perform the work.]
15. Data entry and validation At the completion of each survey day, it should be the responsibility of the team coordinator and delegated helper(s) to re-check all record forms for completeness. At this stage the opportunity may still exist to correct any ambiguities. Plasma glucose values should be transcribed from the laboratory list to individual survey forms if appropriate. (b) The survey form (see Appendix) should require minimal secondary coding and is designed to be entered on-site using a portable computer. [This can be undertaken within 1-2 days of (a) data collection, if data entry clerks are available. If routine biochemistry, such as measurement of lipids, creatinme and uric acid, is being performed at the time of the survey, then these results can also be added to the computer file. This can facilitate the production of computer-pnnted results for survey subjects within a few days of their attendance l
Acknowledgements The authors wish to note the significant contributions of Dr Richard Taylor and Dr Hilary King in the evolution of the survey techniques used by the International Diabetes Institute. We also thank Professor K.G.M.M. Alberti and Professor Jaakko Tuomilehto for their more recent assistance.
SUMMARY Field surveys of diabetes and other noncommunicable diseases and their risk factors are increasingly being performed in both developed and developing countries. Although individual priorities and circumstances will influence survey design, a model survey manual and protocol is described which might assist investigators in formulating plans and preparing manuals for their own surveys. The model presented has been successfully used in developing countries, for both rural and urban communities, over more than a decade. lt describes examples of survey objectives, and gives specific details of survey procedures and measurement techniques which may be utilized. lt contains sufficient flexibility to allow its adaptation to local circumstances.
RESUME Modele de protocole pour I' etude sur le terrain du diabete et des autres maladies non transmissibles Dans les pays developpes comme dans les pays en developpement, les enquiHes sur le terrain portant sur le diabete et des autres maladies non transmissibles et sur les facteurs de risque qui leur sont associes se multiplient. Bien que le plan de ces enquiHes varie en fonction des priorites et des circonstances particulieres, un modele de manuel et de protocole d'enquete a ete etabli afin d'aider les enqueteurs formuler leurs propres plans d'etude et leurs propres manuels. Le modele presente ici est applique avec succes dans des pays en developpement, tant en milieu rural qu'en milieu urbain, depuis plus d'une decennie. 11 donne des exemples d'objectifs d'enquete et decrit en detail les methodes d'enquete et les techniques de mesure pouvant etre utilisees. Ce modele est suffisamment souple pour pouvoir etre adapte aux conditions locales.
a
Rapp tnmest statJSt. santt mond, 45 ( 1992)
-369-
REFERENCES-REFERENCES 1. KING, H. & ZIMMET, P. Trends in the prevalence and incidence of diabetes: non-insulindependent diabetes mellitus. World health statistics quarterly, 41:190-199 (1988). 2. RosE, G. ET AL. Epidemiologic methods in diabetic macrovascular disease. Diabetes care, 2:91-97 (1979). 3. BENNETT, P.H. Recommendations on the standardization of methods and reporting of tests for diabetes and its microvascular complications in epidemiologic studies. Diabetes care, 2:98-104 (1979). 4. RosE, G. ET AL. Cardiovascular survey methods. 2nd edition. WHO Monograph Series No. 56. Geneva, WHO (1982). 5. ZIMMET, P. ET AL. The epidemiology and natural history of NIDDM - lessons from the South Pacific. Diabetes/metabolism reviews, 6:91-124 (1990). 6. COLUNS, V.R. ET AL. An inconsistent relationship between insulin and blood pressure in three Pacific island populations. Journal of clinical epidemiology, 43:1369-1378 (1990). 7. TUOMILEHTO, J. ET AL. ECG abnormalities in relation to glucose tolerance and other risk factors in men of the developing Pacific nation of Kiribati. Nutrition, metabolism and cardiovascular diseases, 1:195-200 (1991). DowsE, G.K. ET AL. High prevalence of NIDDM and impaired glucose tolerance in Indian, Creole and Chinese Mauritians. Diabetes, 39:390-396 (1990). ZIMMET, P.Z. ET AL. The relation of physical activity to cardiovascular disease risk factors in Mauritians. American journal of epidemiology, 134:862-875 (1991). NAN, L. ET AL. Prevalence and medical care of hypertension in four ethnic groups in the newlyindustrialized nation of Mauritius. Journal of· hypertension, 9:859-866 (1991 ). ARMITAGE, P. & BERRY, G. Statistical methods in medical research. 2nd edition. Oxford, Blackwell Scientific Publications (1987). WHO Technical Report Series No. 727, 1985 (Diabetes mellitus: report of a WHO Study Group). OMS, Serie de Rapports techniques No 727, 1985 (Le Diabete sucre: Rapport d'un Groupe d'etude de I'OMS). BURRIN, J.M. & ALBERTI, K.G.M.M. What is blood glucose: can it be measured? Diabetic medicine, 7:199-206 (1990).
8.
9.
10.
11.
12.
13.
Wld hlth statist. quart., 45 (19921
Appendix
-370MAURITIUS NON-COMMUNICABLE DISEASES SURVEY 1992 SURVEY NO: Household No.
Annexe
DEMOGRAPHY
1987 Survey No. (If Applicable)
SURNAME ...................................................................... OTHER NAME(s) .............................................................. .
SEX ADDRESS (Street) ................................................................................................................... (Locality) .................................................................................................................
Male Female 2
DATE OF BIRTH
rnrnrn Day Month Year
Age (Years) Hours Yes 1 No 2 Don't know 3 Years None Diet only Herbal Oral Drug Insulin 1 2 3 4 5
GTTSTATUS
How many hours since last food or drink (except water)?
HAS A DOCTOR EVER TOLD YOU THAT YOU HAVE DIABETES?
rn rn rn D
D
IF YES, HOW MANY YEARS SINCE FIRST DIAGNOSED? PRESENT REGULAR TREATMENT FOR DIABETES
rn D D ·D
(If drugs or insulin, specify type/dose.........................................................................................)
HISTORY
HAS ANYONE IN YOUR FAMILY EVER HAD DIABETES? Father Mother (Tick box if applicable)
D
D
Brother/Sister
D
Children
D
Grandparents
D
Yes 1 No 2 Don't know 3
HAS A DOCTOR EVER TOLD YOU THAT YOU HAVE HIGH BLOOD PRESSURE?
Yes 1 No 2 Don't know 3 Yes No 2 Don't know 3
IF YES, ARE YOU CURRENTLY AND REGULARLY TAKING DRUGS FOR HIGH BLOOD PRESSURE? (Specify.............................................................................................................................) CIGARETTE SMOKING
Never (or < 100 cigs) 1 Ex-smoker (not for ~ 6 mths) 2 Currently < 10/day 3 Currently 10-20/day 4 Currently >20/day 5 Never Ex-drinker (not for ~ 6 mths) Once/Week or less often 2-3 days per week 4 or more dayS/Week Nil .!S.2 per day 3-5 per day > 5 per day Nil Beer Wine/sherry Rum/other spirits Various 1 2 3 4 5 1 2 3 4 1 2 3 4
D D D D D
ALCOHOL (FREQUENCY) How often do you drink?
ALCOHOL (QUANTITY) How much do you drink per day on average? (1 unit chopine of beer or glass of wine or "tor• of spirits)
=
ALCOHOL (TYPE) What kind of alcohol do you most commonly drink?
5
Rapp. trimest. statist. sanit. mond., 45 (19921
-371HAVE YOU EVER BEEN TOLD BY A DOCTOR THAT YOU HAVE, OR HAVE HAD ............ .. Angina Heart attack or myocardial infarction Yes 2 3 No Don't know Stroke Gout Asthma or wheezy bronchitis Peptic (gastric or duodenal) ulcer Specify level and cause .. .... ...... .. .... .... ... ... ...... .... .... .... ........ .... .... ... PHYSICAL ACTIVITY USUAL OCCUPATIONAL PHYSICAL ACTIVITY Lower limb amputation Sedentary (eg office, unemployed) Light (eg sales,sewing,housework) Moderate (eg trades-worker) Heavy (eg labourer) Sedentary (eg house-bound) Light (eg gardening, walking) Moderate (eg aerobic sports 1-2 dayS/Week) Heavy (eg aerobic sports ~ 3 dayS/Week)
1 2 3 4 1 2 3 4
D D D
USUAL LEISURE PHYSICAL ACTIVITY
GENERAL LEVEL OF EDUCATION ACHIEVED None/Standard 1 - 3 Standard 4-6 Form 1-3 Form 4- 6 Tertiary {post H.S.C) 1
2 3 4 5
TOTAL NUMBER OF YEARS FULL-TIME EDUCATION?
Years Professional/Managerial Technical/Clerical Skilled worker/Artisan Partly skilled Unskilled worker Unemployed/Pensioner Housewife
rn D D D
OCCUPATION Specify .............................................................................................. (Previous occupation if unemployed or pensioner, husband's occupation if housewife.)
1 2 3 4 5 6 7
IF HOUSEWIFE, OCCUPATION OF HUSBAND MONTHLY FAMILY INCOME Less than As 2500 As 2500 - 4999 As 5000 - 7500 More than As 7500 Don't know
1 2 3 4 5
OBSTETRIC HISTORY (Women only) ARE YOU PREGNANT NOW? If yes, number of weeks
rn
Yes 1 No 2 Don't know3 Yes 1 No 2 Don't know3 Yes 1 No 2 Don't know 3 Number Number Number Yes No 2 Don't know3
D D D
IF NO, HAVE YOUR PERIODS CEASED FOR THE LAST 6 MONTHS OR MORE? (ie. Have you reached the menopause?)
ARE YOU CURRENTLY USING STEROID CONTRACEPTIVES? (If yes, specify ........................................................................................................................) HOW MANY BABIES HAVE YOU HAD BORN ALIVE? HOW MANY BABIES HAVE YOU HAD BORN DEAD (STILLBIRTHS)? HOW MANY MISCARRIAGES HAVE YOU HAD? HAVE ANY BABIES HAD CONGENITAL MALFORMATIONS? (If yes, specify ........................................................................................................................)
rn rn rn D
Wld hlth stattst. quart, 45 I 1992)
-372-
EXAMINATION ObsD Height (cm) Weight (kg)
I I If I IJI IJI II I If I I I I I
GIRTH
ObsD
1.
Waist (cm) Hip (cm)
3.
Waist (cm) Hip (cm)
;::1
==::===:~~=: ObsD
2.
Waist (cm) Hip (cm)
I.__....___.___.___,
BLOOD PRESSURE
1. Systolic (mmHg) Diastolic 5
Pulse {15secs)
rn
3.
Systolic (mmHg) Diastolic 5
l'r==r==r====.
2. Systolic (mmHg) Diastolic 5
I
FASTING PLASMA GLUCOSE
(mmol/~
TWO HOUR PLASMA GLUCOSE (mmol/1) CHECK
EEB D D 2-hour specimen .................... Time taken Lean body mass
D D NOTES
Fasting specimen
ECG
D D
Glucose load
I
J
Physical activity
Rapp. tnmest. stattst samt. mond, 45 (1992)