Description and quality of baseline data:
Blood pressure
Kari Kuulasmaa1, Matti Niemelä1 and Bijoy Joseph1,2 for the MORGAM Project3
|
Description and quality of baseline data:
|
1 Department of Health Promotion and Chronic Disease Prevention, National Public Health Institute, Helsinki,
Finland
2 Since January 2007 at Indic Society for Education and Development
(INSEED), Nashik, India
3 See Annex for the sites and key personnel of
contributing MORGAM Centres
|
© National Institute for Health and Welfare
and the MORGAM Project investigators Last updated: 4 July 2007 For more information, please contact Kari Kuulasmaa (firstname.lastname@thl.fi) |
MORGAM collected data on blood pressure, which was measured in the baseline examination. The data were transferred to the MDC using the Data transfer format - MONICA survey data (Form 20), which has the following relevant data items:
The two blood pressure readings refer to two subsequent measurements.
For data analysis, three derived variables had been defined by the time of the preparation of this document:
Of these, SYSTM and DIASTM are most commonly used variables in data analysis.
This description and quality assessment consists of:
This is an adaptation and simplification of the MONICA quality assessment, which also included:
Concerning the timing and environment of the measurements, MORGAM has collected data on the time of day of the blood pressure measurements and room temperature during the blood pressure measurements. These were not evaluated systematically in this quality assessment report.
The MPCs were asked to report the type of sphygmomanometer, the side of stethoscope (bell or diaphragm) and the widths of the bladders of the cuffs that were used for blood pressure measurement in the baseline examination of their MORGAM cohorts. The simple mercury sphygmomanometer and the random zero sphygmomanometer, if used correctly, are reliable devices. Those are becoming replaced increasingly by automated devices. However, automated devices vary considerably in quality of the measurement. Two evaluation procedures were introduced for blood pressure measurement devices in 1993, one by the British Hypertension Society (BHS) [1], and another one by the Association for the Advancement of Medical Instrumentation (AAMI) [2]. Because of difficulty in fulfilling the criteria of these evaluation protocols, looser international criteria were introduced by a Working Group on Blood Pressure Monitoring of the European Society of Hypertension in 2002 (International Protocol) [3].
A Device Score (DEV) was defined as
| DEV = | 2 | if
was used; |
| 1 | if automated device which has passed the International Protocol or the BHS validation (class B) was used; | |
| 0 | if the blood pressure measurement device did not fulfil the criteria for 2 or 1. |
A list of validated blood pressure devices is available at http://www.bhsoc.org/blood_pressure_list.stm.
The value of blood pressure measurement depend on the width of the bladder of the cuff: a narrow cuff gives too high values if the arm circumference is large, and a wide cuff gives too low values if the arm circumference is small [ref]. The effect of cuff sizes in population surveys is described in the MONICA quality assessment. The widths of cuffs used in the MORGAM cohorts are described.
The MPCs were also asked to report the blood pressure measurement procedures used in the baseline examination. This included the timing of the blood pressure measurement with respect to the venipuncture for blood sampling, the posture of the subject during blood pressure measurement, the arm (left or right) from which blood pressure was measured, the number of subsequent measurements on each person and the order of the measurements which were reported for MORGAM.
The mean, standard deviation, percentiles, and the extreme values were tabulated for each blood pressure and random zero reading. In addition, three quality indicators were calculated from the data. These were:
A subject's measurement was considered complete only if two measurements of both systolic and diastolic blood pressure were recorded. A Proportion of Incomplete Measurements score (PIM) was defined as:
| PIM = | 2 | if less than 5% of the measurements were incomplete |
| 1 | if 5% or more of the measurements were incomplete |
Note that the definition of PIM differs from that used in MONICA. It recognizes the fact that non-MONICA cohorts were accepted for MORGAM even if they had only one blood pressure measurement from each person.
The terminal digits of the blood pressure readings are expected to be distributed uniformly over {0, 2, 4, 6, 8} when mercury sphygmomanometers were used, and over {0, 1, 2, 3, 4, 5, 6, 7, 8, 9} for automated sphygmomanometers. For manual blood pressure measurements there tends to be zero preference for the terminal digits, and high zero preference is a strong indicator of lack of training of the blood pressure measurer for research measurements or lack of quality control during the blood pressure measurements. The distributions of the terminal digits for the original (i.e. not corrected for random zero) single measurements were tabulated. A Digit Preference Score (DPS) was defined as in MONICA using the formula:
DPS = 100 * ( X2 / df * N )1/2
where N is the number of observations, X2 is the chi-square statistic for the test of homogeneity of the terminal digits, and df is the respective degrees of freedom (for example, df = 4 when a single blood pressure reading is used because there are 5 possible even terminal digits). The DPS ranges from 0 to 100. It is low for high agreement when there is no preference for the terminal digit and rises consistently with the loss of such agreement. DPS was calculated from all single even blood pressure readings. For cohorts where a random zero sphygmomanometer was used, the DPS were calculated from the original readings without correcting for random zero. From the values of DPS, a Terminal Digit Preference score (TDP) was defined as:
| TDP = | 2 | if both systolic and diastolic DPS are less than 10% |
| 1 | if at least one of systolic and diastolic DPS is between 10% and 33% and none reaches 33% | |
| 0 | if systolic or diastolic DPS is 33% or more |
Note that TDP defined above has different cut-points compared to the definition used in MONICA. A TDP value of "0" indicates a severe concern about the validity of the data.
DPS was also calculated for the mean of the two measurements (with df=9), after correction for random zero where relevant.
PIR was defined here using the MONICA definition. After correction for random-zero errors, differences between the first and the second measurements of both systolic and diastolic blood pressure are calculated for each individual and the percentage of identical measurements are calculated. A Proportion of Identical Results in duplicate blood pressure measurements score (PIR) is defined as:
| PIR = | 2 | if less than 33% of the duplicate systolic as well as diastolic BP measurements are identical |
| 1 | if proportion of identical results for either systolic or diastolic blood pressure measurements, is between 33% and 50% and none is above 50% | |
| 0 | if 50% or more of the duplicate systolic or diastolic BP measurements are identical |
Due to natural variation in individual's blood pressure, a very high percentage of identical results in subsequent measurements suggests that the two measurements are not independent but the result of the first measurement has influenced the result of the latter. High zero preference in the terminal digits also increases the proportion of identical results in duplicate measurements. PIR reflects the training of the measurers and quality control of the measurements.
A Summary Score of Quality (SSQ) was defined to summarize the quality from what were considered the most critical quality indicators. It was defined as the minimum of the four scores: DEV, PIM, TDP and PIR, i.e.
SSQ = min{DEV, PIM, TDP, PIR}.
If only one blood pressure reading is available then it is not possible to calculate PIR and, in such cases, SSQ is defined as minimum of the remaining three scores DEV, PIM and TDP. Then the maximal value of SSQ is 1 because PIM will be one.
Table 1 gives the instruments used for blood pressure measurement. Most used the standard mercury sphygmomanometer or the Hawksley random-zero sphygmomanometer. The PRIME cohorts (FRA-LIL, FRA-STR, FRA-TOU and UNK-BEL) used the Spengler SP9 automated blood pressure measurement device. We are unfortunately unaware of any validation of this device. Therefore, these cohorts got DEV score "0". All other cohorts had DEV score "2".
Table 1 also gives the widths of the bladders of the cuffs used in the Cohorts. The use of different cuffs is not available for all Cohorts which reported the use of different cuffs (see the distribution if item CUFF). For the Cohorts for which the frequencies of the use of different cuffs is available, nearly always the same cuff was used. We do not consider the width of cuff a major quality issue for the follow-up analysis in MORGAM.
Table 2 gives the blood pressure measurement procedures used by the MPCs. Most Cohorts reported the first and the second measurement for MORGAM. Cohorts 02, 03 and 21 of DEN-GLO reported the 3rd and 4th measurements, which are likely to be a bit lower and more stable than the 1st and 2nd measurements. For Cohort 01 they reported the 1st and 2nd measurements. FIN-ATB, FRA-LIL, FRA-STR, FRA-TOU, UNK-BEL measured blood pressure only once. UNK-CAE took three measurements, only the first measurement which was taken using the random zero device, is available in MORGAM. The other two measurements were taken for experimental purposes using an automated device.
Here are hyperlinks to the distributions of the data items:
The most striking finding in the distribution tables is that only one blood pressure measurement is available in FIN-ATB, FRA-LIL, FRA-STR, FRA-TOU, UNK-BEL and UNK-CAE. This is consistent with the measurement procedures reported by the MPCs. Furthermore, although the random zero device was used in UNK-CAE, only the corrected values are available for MORGAM.
Table 3 gives the proportion of subjects with incomplete measurements and the values of the PIM scores. The percentage of incomplete measurements is mostly zero. For FIN-ATB, FRA-LIL, FRA-STR, FRA-TOU, UNK-BEL, and UNK-CAE, for which only one blood pressure measurement is available, the percentage of incomplete measurements is 100 and hence PIM = 1.
Table 4 gives the distribution of the terminal digits of the blood pressure readings. It reveals that odd terminal digits were recorded vary rarely except for FRA-LIL, FRA-STR, FRA-TOU and UNK-BEL, which used an automated device.
Table 3 gives DPS as well as TDP. It also gives DPS calculated using mean of the first and second measurements. TDP is not necessarily valid for UNK-CAE, where only the measurements corrected for random zero are available.
It was already observed in MONICA, that proportion of RUAs with TDP = zero decreased in Cohort 03 compared to Cohorts 01 and 02 and the proportion of RUAs with score 2 increased in Cohorts 02 and 03 compared to Cohort 01.
Table 3 gives the percentage of identical readings of the two measurements and the values of the PIR scores. PIR score is mostly 2, but 1 or 0 for some Cohorts. It was not possible to calculate PIR for six cohorts for which only one blood pressure measurement was available.
The values of SSQ are given in Table 3. The summary score seem to improve steadily from Cohorts 01 to Cohorts 02 and Cohorts 03. For nearly all of the non-MONICA Cohorts, SSQ is either 1 or 0. For four cohorts the summary score is zero due to the use of an non-validated automated device (FRA-LIL, FRA-STR, FRA-TOU, UNK-BEL). For few other cohorts the terminal digit preference score is zero suggesting insufficient training of the blood pressure measurers (ITA-ROM, Cohorts 21 and 22, POL-TAR Cohort 01, RUS-NOV Cohort 02, UNK-GLA Cohort 21).
The following list includes only the RUAs with specific findings or exceptional background information relevant for the use of the data:
DEN-GLO
FIN-ATB
FRA-LIL, FRA-STR and FRA-TOU
ITA-ROM
POL-TAR
POL-WAR
RUS-NOV
UNK-BEL
UNK-CAE
UNK-GLA
| Date | Update |
|---|---|
| 2007-07-04 | First published version. |