Comparison of blood pressure measured at the arm, ankle and calf


Dr A. Dobson


The suitability of alternative sites for non-invasive blood pressure (NIBP) measurement was investigated in 100 awake healthy volunteers. The calf and the ankle were chosen for comparison with the arm, and the results analysed subjected to Bland–Altman analysis. Discomfort was graded using a Visual Analogue Scale. There was a poor agreement between the different sites with respect to systolic blood pressure: the agreement was closer for diastolic and mean measurements. The mean blood pressure calf measurement was on average 4 mmHg (95% limits of agreement −12 to 20), higher than the arm. The ankle was 8 mmHg higher (−8 to 24) than the arm. ANOVA demonstrated a statistically significant difference in the discomfort scores between the sites (p < 0.001). The calf demonstrated the highest discomfort score and the ankle the lowest. We suggest that the ankle should be considered in preference to the calf as an alternative site for NIBP measurement if use of an arm is undesirable or impossible.

It is well established and standard practice to measure non-invasive blood pressure (NIBP) in the peri-operative setting using oscillotonometric equipment. The most common site of blood pressure measurement is the upper arm. However there are times when it is either not ideal or not possible to use the upper limb for blood pressure measurement.

This is particularly relevant in surgery where only one upper limb is accessible. In these circumstances, particularly when TIVA is being administered in the same arm, having the cuff inflating regularly may impede the smooth delivery of anaesthetic. There are also occasions, particularly in orthopaedic or plastic surgery, where neither arm is available, and in these circumstances an alternative site has to be used for BP measurement.

We decided to investigate the suitability of other sites for non-invasive blood pressure measurement. Previous studies that have examined alternative sites for BP measurement have either used a limited patient population or have not used a Bland–Altman approach for analysis [1, 2], which is widely considered to be the most appropriate way of comparing two methods of measuring a physiological variable. We choose to study both adult male and female volunteers.

The sites we chose to look at were the calf and the ankle. The calf, in our experience, is the site most commonly chosen for NIBP measurement when the arm is not available. This has potential disadvantages due to the large muscle bulk leading to discomfort for the conscious patient prior to induction of anaesthesia. Utilisation of the posterior tibial artery at the ankle may be preferable to the calf, due to ease of access and similar circumference to the upper arm. Also the lack of muscle bulk may lead to less discomfort on cuff inflation.

We proposed to use a standard automated NIBP device to compare measurements made at the arm, calf and ankle. We anticipated that BP measurement would be higher in the lower limbs, as our vascular surgical colleagues use this fact to help them determine the presence or absence of peripheral vascular disease. In most people the resting ankle pressure is greater then the pressure at the arm – this is known as the ankle-brachial index (ABI). The norm for this value is 1.1.

We hoped to demonstrate a predictable correlation between arm/ankle and arm/calf BP readings in our volunteers. If we could also demonstrate lower discomfort scores when using the ankle as opposed to the calf, then we would be able to recommend BP measurement at the ankle in preference to the calf when the arm is unavailable.


We applied for and received approval from the South Manchester Research Ethics Committee for our study (ref 05/Q1403/209). We also registered with the Pan-Manchester Research and Development Department.

One hundred volunteers were used for the study. These were members of staff that worked in the theatre environment and varied from Medical staff, Operating Department Practitioners and Nurses and students. The 100 volunteers were made up of 65 women and 35 men with an age distribution between 20 and 64 years. Informed consent was obtained for participation in the study.

Inclusion/exclusion criteria were

  • • Adults of either sex.
  • • No major heart, lung or vascular disease or any other disease that affects routine daily activities.
  • • Not already known to be hypertensive or on treatment for it.
  • • No skin lesions over arm, calf, or ankle that would prevent measurements.
  • • Pregnant volunteers were excluded.

Each volunteer lay on a trolley in a semi-recumbent position (30° head up tilt) and had their BP measured at each of the three sites – arm, calf and ankle on both sides in a predetermined random order ( This was confirmed using chi-square tests (all p values > 0.05).

For measurement of NIBP the same Datex S5 series monitoring machine (Datex HQ, 1200 West 7th Street, Los Angeles, CA 90017, USA) was used for all of the volunteers. Prior to data collection this was checked and calibrated by the Medical Physics Department and found to be within the accepted range of variability. The machine was not used in any other clinical situations during the period of data collection.

When the Datex NIBP is switched on/reset it performs a preset measurement of BP starting at 185 mmHg cuff inflation pressure. It then has a memory capability which allows it to perform further measurements closer to the systolic pressure determined from the original measurement, in an effort to reduce patient discomfort. For our purposes, the machine was reset between measurements so that the machine treated each individual measurement as though it were a new patient, starting at the preset level of 185 mmHg.

At each site an appropriately sized Datex BP cuff was used. Small, standard and large adult cuffs were fitted, the bladder of the cuff being at least 40% of the circumference of the limb being measured. According to standard guidelines the middle of the bladder was placed over the artery being measured.

For the upper arm this was in the standard position. For the calf the cuff was placed over the mid aspect of the posterior calf (where one would anticipate the popliteal artery to be). A large adult cuff was found to be an appropriate size for most volunteers (Fig. 1).

Figure 1.

 Cuff placement on calf.

For the ankle often a small or standard adult cuff was appropriate. This was placed in the most distal position possible of the lower limb with the mid bladder placed just posterior to the medial malleolus in order to measure pulsations from the posterior tibial artery (Fig. 2).

Figure 2.

 Cuff placement at ankle.

After each measurement the volunteer was asked to grade the discomfort they felt during the measurement on a Visual Analogue Scale (VAS) from 1 to 10: 1 meaning no discomfort; 10 meaning severe discomfort.

Results were documented manually and then entered into an excel database and prepared for statistical analysis. Prior consultation with the medical statisticians had given us a power calculation of 100 volunteers required to achieve statistical significance. With a sample size of 100, the study has 80% power to detect differences in VAS of 1 or more between the sites, assuming an expected SD of within subject differences of 2 and an overall 5% level of significance adjusted for multiple comparisons between the three sites. This sample size would also enable 95% limits of agreement to be calculated with associated confidence limits of ±0.34 SD of measurement differences. We planned to analyse the results in terms of discomfort score, and then each of the systolic, diastolic and mean BP separately in order to make interpretation easier. We used ANOVA to determine if any relationship existed between the different sites. Bland–Altman analysis was used to compare the BP readings at the different sites.


Data collected was complete for 98 out of the 100 volunteers. On two occasions the volunteers found the calf BP measurement too painful to proceed.

Blood pressure results

The results were analysed comparing systolic, mean and diastolic pressures at each of the three sites. Each blood pressure was measured at six locations in the 100 volunteers.

Limits of agreement

The agreement between the blood pressure readings was assessed by using a Bland–Altman approach. The calf and ankle readings were compared with the arm reading which was taken as the ‘gold’ standard. The measurements used were the average of the readings taken on the right and left sides. The limits of agreement were calculated by evaluating the difference between each pair of scores (Table 1).

Table 1.   Limits of agreement for systolic, diastolic and mean BP (mmHg).
 Mean difference95% limits of agreement
Systolic BP
 Calf – arm22.4 −2.7 to 47.4
 Ankle – arm17.8 −5.5 to 41.1
Diastolic BP
 Calf – arm−3.7−21.2 to 13.9
 Ankle – arm2.7−14.0 to 19.4
Mean BP
 Calf – arm3.7−12.3 to 19.7
 Ankle – arm7.9 −8.2 to 24.0

An example of how the limits of agreement values can be interpreted is as follows. With respect to the mean BP values, for an individual the calf measurement is on average about 13 mmHg higher than the arm measurement. However, this difference could be as much as 3 mmHg lower or up to 28 mmHg higher. Similarly the results for the ankle and arm imply the following; for an individual the ankle measurement is, on average, about 17 mmHg higher than the arm measurement, but could be as much as 33 mmHg higher.

The results for diastolic pressure show a higher level of agreement than those results for the systolic pressures.

The mean blood pressures show better agreement than systolic readings but not as much as the diastolic readings.

Discomfort Score

Each of the 100 volunteer patients scored the discomfort of the blood pressure test (1–10) at each of the six locations.

A square root transformation was used to normalise the data. The transformed discomfort scores were than analysed using repeated measures ANOVA (Table 2).

Table 2.   Marginal means of discomfort scores.
TermMean95% confidence interval
  1. All transformed values have been squared.

 Right side3.02.7–3.3
 Left side3.12.8–3.4
Interaction term
 Right arm2.52.2–2.8
 Left arm2.72.3–3.0
 Right calf5.44.9–5.9
 Left calf5.34.8–5.8
 Right ankle1.71.5–1.9
 Left ankle1.91.6–2.1

The results from the ANOVA show that there is a statistically significant difference in the discomfort scores between the sites (p < 0.001). Pairwise comparisons between the sites, adjusted for multiple comparisons, indicate that each of the sites is different (p < 0.001).

The calf has the highest discomfort score and the ankle has the lowest.

The ANOVA shows a borderline significant difference between the sides (p = 0.07) i.e. the left side has a slightly higher discomfort score than the right side although the actual difference was very small.

Association between blood pressure values and order of measurement

In order to investigate whether the order in which the readings were taken could influence subsequent readings we looked at the association between the blood pressure values and the order of measurements using Spearman correlations (Table 3).

Table 3.   Correlation between blood pressure values and order of measurement.
Corr. coeff.p valueCorr. coeff.p valueCorr. coeff.p value
  1. Unadjusted p values are shown.

Left arm−0.1160.2520.0370.714−0.0910.366100
Right arm0.0390.703−0.0060.9490.0300.766100
Left ankle−0.0070.947−0.0550.5870.0260.799100
Right ankle−0.1140.257−0.0960.340−0.0780.441100
Left calf−0.3020.002−0.1600.115−0.1500.13999
Right calf−0.2270.024−0.0620.543−0.2090.03998

There is no evidence of an association between the systolic or diastolic blood pressure readings and the order in which the measurements were taken at the arm or ankle sites. There is evidence however, of a small negative correlation in systolic blood pressure with the order the measurements were taken for both left and right calves. This association indicates that calf measurements have a wider variability over the repeats, in that measurements made in ‘earlier’ orders tend to be higher than those made in ‘late’ orders. However, the analysis on the differences in blood pressure measurements between sites should not be biased because the order of measurements is random.


A number of previous studies have looked at alternative sites for measurement of NIBP. They have found conflicting results. Block and Shulte [1] concluded that ankle placement of BP cuff was an acceptable alternative to arm measurement. This study involved 24 patients undergoing general anaesthesia. Sanghera and North [2] compared arm and ankle blood pressure under spinal anaesthesia for Caesarean section. They found only marginal agreement between the two methods and concluded that the ankle was not a suitable alternative in this circumstance. Only the latter of these studies used the Bland–Altman method of analysis. Bland and Altman [3] is widely recognised as being the most appropriate way of comparing two methods of measuring a physiological variable.

However, this study by Sanghera and North [2] was looking only at the obstetric population. We felt that by investigating the relationship between NIBP in a non-anaesthetised group of essentially normal individuals we may be able to draw more wide reaching and relevant conclusions from the data.

We measured NIBP in non-anaesthetised individuals to see if there was a degree of agreement between arm, calf and ankle and also to determine the level of discomfort involved at an alternative site. We employed the Bland–Altman approach to our analysis.

Our findings show that the ankle is the most comfortable site for measuring NIBP and that the calf is by far the most uncomfortable. There may be a number of reasons for this. The main one, we suspect, is related to the amount of muscle mass beneath the cuff.

In terms of actual agreement of BP at each of the three sites we found a much closer agreement between diastolic BP than systolic BP. This implies that the mean arterial pressure was somewhere between the two. Diastolic pressure has much less clinical significance to us as clinicians than systolic or mean BP.

A potential valid criticism of the study design is that the discomfort caused by inflation of the cuff at the calf may have led to elevated subsequent readings at the other sites investigated. Previous studies measuring NIBP at multiple sites have managed to synchronously measure BP [4]. Due to the multiple number of sites involved in our study, this approach would have not been practicable or possible. Re-setting of the NIBP machine between each reading in our study allowed the subject time to compose him/herself before subsequent readings were taken (usually around 1–2 min). Spearman correlation calculations confirmed our suspicion that discomfort at the calf had an increasing effect on subsequent BP values. Measurements involving ankle or arm at the beginning of the sequence had no such effect.

In conclusion our results suggest that when using the calf for NIBP measurement that significantly higher readings as compared to the arm may be encountered and this cofirms the results of other studies. This must be taken into account whenever the lower limb is the only available site. In view of the fact that, in comparison with the arm, the readings we obtained at the calf and the ankle were similarly elevated, but the discomfort scores were significantly higher for measurements obtained at the calf, we would suggest that the ankle should be considered preferable to the calf, particularly in anxious patients in whom readings are to be taken prior to induction of anaesthesia.

It is the author’s (AD) personal practice, when anaesthetising patients in whom per-operative access to the arms is problematic, to take a single blood pressure reading at the arm before proceeding to use the ankle for ongoing NIBP measurement. This gives an indication for a given patient of the degree of difference between the two sites, and allows appropriate interpretation of subsequent results.

In conclusion, NIBP readings taken from the lower limb, either at the calf or at the ankle, tend to be higher when compared with the arm. NIBP readings taken at the ankle are associated with significantly less discomfort than those taken at the calf.


The authors wish to express their hanks to Howard Thompson, Department of Medical Physics, South Manchester University Hospitals NHS Trust and to Sr. Jane Curtis, Stockport NHS Foundation Trust.

Conflicts of interest

None to declare.