WWW-publications from the WHO MONICA Project
October 1998
Anu Molarius1, Susana Sans2 and Kari Kuulasmaa1 for the WHO MONICA Project3
1 MONICA Data Centre, National Public Health Institute, Helsinki, Finland
2 Department of Health and Social Security, Institute of Health Studies,
Barcelona, Spain
3 Annex: Sites and key personnel of the WHO MONICA
Project
Thanks are due to Hermann Wolf for his advice during the preparation of an earlier quality assessment report on the topic and to Alun Evans who commented on the text.
The MONICA Centres are funded predominantly by regional and national governments, research councils, and research charities. Coordination is the responsibility of the World Health Organization (WHO), assisted by local fund raising for congresses and workshops. WHO also supports the MONICA Data Centre (MDC) in Helsinki. Not covered by this general description is the ongoing generous support of the MDC by the National Public Health Institute of Finland, and a contribution to WHO from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland, USA for support of the MDC. The completion of the MONICA Project is generously assisted through a Concerted Action Grant from the European Community. Likewise appreciated are grants from ASTRA Hässle AB, Sweden, Hoechst AG, Germany, Hoffmann-La Roche AG, Switzerland, the Institut de Recherches Internationales Servier (IRIS), France, and Merck & Co. Inc., New Jersey, USA, to support data analysis and preparation of publications.
Waist and hip circumference measurements were introduced in the November 1990 version of the MONICA Manual (3). Since most of the MONICA Collaborating Centres (MCCs) had already started and many had even finished their middle population survey by that time, waist and hip circumferences were considered as optional measurements in this survey. As a result, only about half of the MCCs made these measurements in their middle surveys. In the final survey, waist and hip circumferences were measured in almost all MCCs.
If waist and hip circumferences are measured following standard procedures and the observers have been properly trained, the measurements should be reasonably reliable. Otherwise, the measurements are vulnerable to bias. Potential sources of bias include incorrect anatomical level of measurement, too heavy clothing or incorrect position of the subject and applying inappropriate tension to the tape measure.
Waist and hip measurements are mainly used for the waist-hip ratio (calculated as waist (cm)/hip (cm)). Waist-hip ratio is generally considered as the best measure of abdominal obesity, which has been suggested as being more closely linked with the risk of cardiovascular disease than other forms of obesity.
Table A shows the bias (%) in waist-hip ratio induced by different biases in waist and hip measurements. The table illustrates that if the biases in waist and hip readings are proportionally the same and in the same direction, there is no bias in waist-hip ratio. Because waist and hip are measured with the same device and by the same observer, generally, this could be the case in most situations where bias exists (e.g. in populations which have measured waist and hip circumferences in full centimetres and both waist and hip are rounded to the full cm in the same way). However, if the bias is introduced in only one of these measures, for instance by measuring waist or hip circumferences at wrong level, or if the biases in waist and hip go in opposite directions, the bias in waist-hip ratio can be substantial. Since the standard error of the mean waist-hip ratio within a 10-year age group of a MONICA sample is around 0.3-0.5%, a bias of 1% can by itself produce a statistically significant difference between two populations.
| Bias in hip | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (%) | -10.0 | -5.0 | -2.0 | -1.0 | -0.5 | 0.0 | 0.5 | 1.0 | 2.0 | 5.0 | 10.0 | |
| Bias in waist |
-10.0 | 0.0 | -5.3 | -8.2 | -9.1 | -9.5 | -10.0 | -10.0 | -11.0 | -12.0 | -14.0 | -18.0 |
| -5.0 | 5.6 | 0.0 | -3.1 | -4.0 | -4.5 | -5.0 | -5.5 | -5.9 | -6.9 | -9.5 | -14.0 | |
| -2.0 | 8.9 | 3.2 | 0.0 | -1.0 | -1.5 | -2.0 | -2.5 | -3.0 | -3.9 | -6.7 | -11.0 | |
| -1.0 | 10.0 | 4.2 | 1.0 | 0.0 | -0.5 | -1.0 | -1.5 | -2.0 | -2.9 | -5.7 | -10.0 | |
| -0.5 | 10.6 | 4.7 | 1.5 | 0.5 | 0.0 | -0.5 | -1.0 | -1.5 | -2.5 | -5.2 | -9.5 | |
| 0.0 | 11.1 | 5.3 | 2.0 | 1.0 | 0.5 | 0.0 | -0.5 | -1.0 | -2.0 | -4.8 | -9.1 | |
| 0.5 | 11.7 | 5.8 | 2.6 | 1.5 | 1.0 | 0.5 | 0.0 | -0.5 | -1.5 | -4.3 | -8.6 | |
| 1.0 | 12.2 | 6.3 | 3.1 | 2.0 | 1.5 | 1.0 | 0.5 | 0.0 | -1.0 | -3.8 | -8.2 | |
| 2.0 | 13.3 | 7.4 | 4.1 | 3.0 | 2.5 | 2.0 | 1.5 | 1.0 | 0.0 | -2.9 | -7.3 | |
| 5.0 | 16.7 | 10.5 | 7.1 | 6.1 | 5.5 | 5.0 | 4.5 | 4.0 | 2.9 | 0.0 | -4.5 | |
| 10.0 | 22.2 | 15.8 | 12.2 | 11.1 | 10.6 | 10.0 | 9.5 | 8.9 | 7.8 | 4.8 | 0.0 | |
Formula used for the table: If bias in waist is 100a%, bias in hip is 100b% and bias in waist-hip ratio (WHR) is 100c%, then
(1+a)waist/(1+b)hip=(1+c)WHR=(1+c)waist/hip,
and hence
c=(1+a)/(1+b) - 1.
The quality assessment of data on waist and hip circumferences in the second (middle) MONICA survey was distributed in May 1996 (MNM 300A, (4)). This report repeats the main findings of MNM 300A and assesses the quality of the data collected in the final survey. It will not be possible to assess the actual biases in the measurements. The focus will be on the procedures used by the MCCs to fulfil the MONICA standardization criteria and on the quality of the data available in MONICA Data Centre (MDC). These will be used as indicators of the performance of the measurers. If standard procedures were used and the data look good, it is likely that the measurements are also unbiased. On the other hand, if standard procedures were not followed and the data look strange, there is a high risk that the data are biased.
The results of this document are reported by Reporting Unit Aggregates (RUAs) which are potential units of analyses of the MONICA data. The RUAs are identified by an abbreviation and a version letter. The abbreviations and reporting units of the RUAs which have collected data on waist and hip circumferences at least in one survey are listed in Table 1.
In the middle survey 22 RUAs measured waist and hip circumferences, and 34 in the final survey. In GER-HAC and GER-KMS in the middle survey, waist and hip were measured in only some of the RUs where the survey was done. In the final survey, four RUAs (GER-BRE, RUS-NOC, RUS-NOI and USA-STA) did not measure waist and hip circumferences.
Information on the actual survey procedures applied by the MCCs were collected in 1991 using the Questionnaire on MONICA Population Survey Procedures (Form VI). The Survey Procedures Questionnaire was further checked concerning the final survey by the MCCs in 1995. The quality assessment of the data provided is based on the data currently available in the MDC.
For the quality analyses all observations within the age group 25-64 were used, except in AUS-NEW, FRA-LIL, FRA-STR, FRA-TOU, LTU-KAU, NEZ-AUC, POL-TAR, POL-WAR, RUS-MOC, RUS-MOI and SWI-TIC where the age range studied was 35-64. Age was defined as age in full years on the date of examination (see DEF1 in Reference (1)). No age or sex adjustment was applied to the data.
The MONICA Manual (November 1990, (3)) gives the following instructions to measure waist circumference:
"The measurement should be recorded at the level midway between the lower rib margin and the iliac crest, rounded to the nearest .0 or .5 cm."
"The circumference should be measured on subjects without heavy outer garments in standing position. The contents of all pockets must be removed. All tight clothing, including the belt, must be loosened. The participant should stand with the feet fairly close together (about 12-15 cm) with weight equally distributed on each leg. Participants should be asked to breath out gently at the time of the reading of the measurement to prevent them from contracting their muscles or from holding their breath. The tape should be held firmly in a horizontal position. It is recommended that the observer is sitting by the participant while taking the measurement."
The March 1992 version of the MONICA Manual (MNM 214, (2)) gives some further instructions for the measurement of waist circumference and the procedures for applying the tape measure. The Manual advised the measuring of waist and hip with the subjects wearing underwear, if possible. It also stipulated checking the length of the tape measure before starting the survey and at least once a month during the survey, and to use a tape with the two sides differently coloured or having a scale only on one side. The MCCs which had collected these data in the past using different techniques were asked not to change their techniques. Many of the MCCs had, however, started their final survey before these additional instructions were available.
For measuring the hip circumference the MONICA Manual gives the following instruction: "Record measurement of maximum circumference over the buttocks, in centimetres to the nearest .0 or .5 cm."
The results from the Questionnaire on MONICA Population Survey Procedures (Form VI) on the procedures applied in the MCCs for measurements of waist and hip circumferences are summarized in Table 2.
Measurement as in MONICA Manual: Apart from CZE-CZE in the middle survey all RUAs reported that they measured waist and hip circumferences as described in the MONICA Manual. CZE-CZE measured waist circumference at the umbilical level in the middle survey.
Number of team members involved in waist and hip measurement: The reported number of team members involved varied between 1 and 100 in the middle survey and between 1 and 36 in the final survey. For most RUAs there were fewer than 10 team members involved.
Observers trained/certified/tested/recertified: In all RUAs the observers were trained, but they were certified or tested during the survey only in 16 RUAs in the middle survey and in 23 RUAs in the final survey. In SPA-CAT the observers were monitored during the survey by a supervisor. In none of the RUAs were the observers recertified during the survey.
Observer code included in the survey form: CHN-BEI, CZE-CZE, GER-COT, GER-ERF, GER-HAC and GER-KMS in the middle survey and RUS-MOC and RUS-MOI in the final survey reported that no observer code was included in the survey form. AUS-PER reported that the observer code was the same as for blood pressure measurements. DEN-GLO reported that the observer was otherwise identifiable.
Removal of heavy outer/other garments: In all RUAs heavy outer garments were removed for the measurement of waist and hip as instructed in the MONICA Manual. Thirteen of the 22 RUAs in the middle survey and 20 of the 34 RUAs in the final survey removed also other garments. For BEL-CHA, BEL-GHE, CZE-CZE, ITA-BRI, ITA-FRI it is known that the subjects were measured in their underwear and for the rest of the RUAs which have removed other garments it is likely that this was also the case. In SPA-CAT the subjects were measured wearing their underwear and a standard light coat made of paper.
Accuracy of measurement: Six of the RUAs (AUS-NEW, AUS-PER, CHN-BEI, DEN-GLO, SWE-GOT and YUG-NOS) reported that they measured waist and hip in full centimetres in the middle survey, the rest in half centimetres as recommended in the MONICA Manual. In the final survey the same RUAs and two more (CAN-HAL and NEZ-AUC) reported that they measured waist and hip in full centimetres.
Number of measurements per subject: Apart from AUS-PER and SPA-CAT which measured waist and hip twice per subject, all RUAs used only one measurement per subject in the middle survey. In the final survey, SPA-CAT and AUS-NEW measured these items twice per subject. AUS-PER (Mid) and SPA-CAT have sent the mean of the two measurements to the MDC.
In all RUAs the instructions for the measurement of waist and hip circumferences were available in the local language. There was no question about the tape measure used or the frequency of checking it in the Survey Procedures Questionnaire. One MCC commented though that in pratice it is difficult to hold the tape in a horizontal position.
When the information given by the MCCs on the survey procedures used was compared to the actual measurements submitted to the MDC several discrepancies were observed. When comparing Table 2 to Table 3, it was noted that GER-AUR and GER-AUU reported that an observer code was included in the survey form in the middle survey, but the data have not been sent to the MDC. CHN-BEI (Mid), RUS-MOC and RUS-MOI reported that an observer code was not included, but the data are available in the MDC. When comparing Table 2 to Table 6, it was noted also that the actual number of observers differed from the reported one in several RUAs. There were also some discrepancies concerning the reported and actual precision of the measurements. These will be discussed in Section 4.
The MDC checks all population survey core data received from the MCCs at the time they are included in the MONICA database. All possible inconsistencies in the data are reported to the MCC to enable correction of errors. The following constraints concern waist and hip measurements.
All violations of these constraints were reported to the MCC for their correction or confirmation. Data values outside the constraint limits were acceptable, but the MCC had to check that the values were not unusual due to data errors. The MCCs were only asked to correct values if they knew that they were incorrect. The currently unresolved constraint violations concerning data on waist and hip circumferences are listed by survey in Appendix 1. There are only a few unresolved constraint violations.
Table 3 gives the number of observations (respondents), proportion of missing data among respondents, and the mean value and the standard deviation of the waist and hip measurements.
Missing data: In the middle survey, except for some RUAs of MONICA East Germany, missing data were not a problem: less than 5% of missing values for waist and hip were observed. In GER-COT, GER-ERF and GER-HAC the proportion of missing data was outstandingly high (79%, 90% and 23% respectively) in the reporting units which have reported to have measured waist and hip. Observer code (WHCODER) was not available for CZE-CZE, GER-AUR, GER-AUU, GER-COT, GER-ERF, GER-HAC and GER-KMS.
In the final survey, four RUAs (BEL-CHA, BEL-GHE, CAN-HAL and YUG-NOS) had more than 10% of missing data, ranging from 18% to 39%. In BEL-CHA and BEL-GHE this is due to the fact that the smoking questionnaire was completed during a home visit making the subjects respondents but waist and hip were measured during a clinical examination which a large part of the respondents failed to attend. Observer code (WHCODER) was not available for SWI-TIC and SWI-VAF.
Mean values and standard deviations: Very exceptional mean values and standard errors could reflect quality problems. There were no strikingly exceptional values, except for mean hip in DEN-GLO in the middle survey which appears surprisingly low when compared to the mean waist in the RUA and to the mean hip circumference in the other RUAs. This was checked by the MCC after MNM 300A, but no indication of a measurement error could be found.
There were, however, substantial changes in the mean values between the middle and final survey in AUS-NEW (3 cm increase in hip), AUS-PER (5 cm increase in waist, 4 cm in hip), CZE-CZE (4 cm decrease in waist, 3 cm in hip), DEN-GLO (3 cm increase in hip while waist decreased), FRA-TOU (7 cm decrease in waist, 4 cm in hip), SWE-NSW (3 cm increase in waist) and YUG-NOS (3 cm increase in waist, 5 cm in hip). In two of these seven RUAs there is a natural explanation: in FRA-TOU the middle survey included only men whereas the final survey included both men and women, and in CZE-CZE waist was measured at the level of umbilicus in the middle survey but the measurement level was standard in the final survey. The other RUAs were asked to check whether these changes were real, e.g. in concordance with the 5-year changes in weight and BMI. This was to exclude the possibility that these changes could be due to error or changes in the measurement techniques.
Tables 4a and 4b give the distributions of the terminal digits for waist circumference in decimals of centimetres and in full centimetres respectively. When comparing Table 4a with Table 2 several discrepancies between the reported and the actual precision of waist measurement were observed. CHN-BEI in the middle survey and CAN-HAL in the final survey have obviously recorded waist to the nearest 0.5 cm and not in full centimetres as they have reported. CZE-CZE, FIN-KUO, GER-AUR, GER-AUU, GER-COT, GER-ERF, GER-HAC, GER-KMSe, ITA-FRI and UNK-GLA in the middle survey and BEL-CHA, BEL-GHE, FIN-KUO, FRA-STRa, GER-AUR, GER-AUU, GER-EGEb, POL-TAR, POL-WAR, RUS-MOC, RUS-MOI, SWE-NSW and UNK-BEL in the final survey have more than 70% of zero values even though they have reported to have recorded waist to the nearest 0.5 cm.
If the measurements are not done properly, there can also be preference of full tens of centimetres. Therefore, it was decided to look also at the distribution of the second last digit (i.e. the terminal digits of full centimetres). If the terminal digits are uniformly distributed, about 10% of the readings have zero as the last digit. Taking into account the standard error of this expected value (about 0.9% for sample size of 1000), the proportion of the terminal zeros would by chance be very rarely more than 12%, and extremely rarely more than 13%. Table 4b shows only one strikingly high proportion of zero digits (RUS-MOI 19.5% in the final survey). In GER-KMS in the middle survey the proportion was also more than 13%.
Tables 5a and 5b give the distributions of the terminal digits for hip circumference in decimals of centimetres and in full centimetres respectively. When comparing Table 5a with Table 2 several discrepancies between the reported and actual precision of the hip measurement were observed. As for waist, CHN-BEI in the middle survey and CAN-HAL in the final survey have obviously recorded hip to the nearest 0.5 cm and not in full centimetres as they have reported. CZE-CZE, FIN-KUO, FIN-NKA, GER-AUR, GER-AUU, GER-COT, GER-ERF, GER-HAC, ITA-FRI and UNK-GLA in the middle survey and BEL-CHA, BEL-GHE, GER-AUR, POL-TAR, POL-WAR, RUS-MOC, RUS-MOI and UNK-BEL in the final survey have more than 70% of zero values even though they have reported to have recorded hip to the nearest 0.5 cm.
Corresponding to the waist measurement, Table 5b gives the distributions of the second last digit in hip circumference. Remarkably high proportions of zero digits were observed in ITA-BRI (16.5%) in the middle survey and in RUS-MOI (15.9%) in the final survey.
When looking at the distributions of the terminal digits in Tables 4a and 5a, it was noted that the zero preferences in waist and hip measurements are very similar, especially when they are over 70%. This suggests that the proportion of zero values might differ between observers within one RUA i.e. some observers may have measured waist and hip in full centimetres whereas others in 0.5 cm. If this is the case, this might be an indication of non-uniform training of observers.
Table 6 gives the number of observers which have a zero preference (<30% or >70%) in last digits of waist and hip measurements, the total number of observers with known observer code, proportion of measurements made by observers with zero preference and the total number of observations made by all observers with known observer code. The table shows that in the RUAs for which the observer code is available the proportion of measurements made by observers with zero preference is usually either very low or very high indicating that the training was reasonably uniform. However, in AUS-PER, CHN-BEI, FIN-KUO, FIN-NKA, FIN-TUL and UNK-GLA in the middle survey and in BEL-GHE, CZE-CZE, FIN-KUO, FIN-NKA, FIN-TUL, FRA-LIL, FRA-STR, GER-AUR, GER-AUU, GER-EGEb, ICE-ICE, POL-TAR, SWE-NSW, UNK-BEL and UNK-GLA in the final survey a difference in zero preference between observers was observed.
Table 7 gives scores for the quality of the data on waist and hip measurements in the middle and final surveys and summary scores for both measurements to be applied for waist-hip ratio. WAIST SCORE and HIP SCORE were defined as:
| Score = | 2 (no indication of a problem) | if the proportion of terminal zeros is between 30% and 70% AND there are <= 13% zeros in the second last digit AND the proportion of missing data is <=30%; |
| 1 (some concern) | if not 2 or 0; | |
| 0 (major concern) | if there are >14% zeros in the second last digit OR the proportion of missing data is >30%. |
The score was also '1' when the measurement level was not standard but still clearly defined (waist in CZE-CZE in the middle survey). This exceptional level of measurement should be mentioned in publications using the data.
Section 4.3 suggested cut-off points 12% and 13% for the proportion of zeros in the second last digit. The cut-off points 13% and 14% used in the scores allow slight deviations from the uniform distribution and are more reasonable for RUAs where the sample size was less than 1000 subjects. The analysis of last digits by observer was not taken into account in the scores because all RUAs did not provide observer codes.
In the middle survey, two RUAs had a zero score for waist: GER-COT had 79% of missing data among respondents, and GER-ERF had 90%. For hip, three RUAs had a zero score: ITA-BRI had more than 14% zeros in the second last digit, and GER-COT and GER-ERF for the same reason as for waist.
In the final survey, only two RUAs had a zero score and both for waist and hip: BEL-CHA had 39% of missing data among respondents (this has been explained in Section 4.2) and RUS-MOI had more than 14% zeros in the second last digit.
The SUMMARY SCORE was derived using the sum of waist and hip scores:
| Score = | 2 | if the sum is 3 or 4; |
| 1 | if WAIST SCORE=1 and HIP SCORE=1; | |
| 0 | if WAIST SCORE=0 or HIP SCORE=0. |
Three RUAs had the summary score '0' in the middle survey (GER-COT, GER-ERF and ITA-BRI) and two RUAs in the final survey (BEL-CHA and RUS-MOI). If the score is zero, there are major reservations concerning the quality of waist and hip measurements and therefore the use of the data.
In addition, even though it is unlikely that the data on waist and hip circumferences will be used in trend analyses, the RUAs which had large unexplained increases or decreases in these measurements between the middle and final surveys (DEN-GLO and YUG-NOS) should be excluded from trend analyses as it is not possible to conclude with certainty that the changes were not due to error or changes in measurement techniques.
The following list includes only the RUAs with specific findings or exceptional background information relevant for the use of the data.
AUS-NEW
AUS-PER
BEL-CHA and BEL-GHE
CAN-HAL
CHN-BEI
CZE-CZE
DEN-GLO
FIN-KUO
FIN-NKA
FIN-TUL
FRA-LIL
FRA-STR
FRA-TOU
GER-AUR and GER-AUU
GER-COT
GER-EGEb
GER-ERF
GER-HAC
GER-KMS
ICE-ICE
ITA-BRI
ITA-FRI
LTU-KAU
NEZ-AUC
POL-TAR
POL-WAR
RUS-MOC
RUS-MOI
SPA-CAT
SWE-GOT
SWE-NSW
SWI-TIC and SWI-VAF
UNK-BEL
UNK-GLA
YUG-NOS