The association of pre-operative home accelerometry with cardiopulmonary exercise variables

Anaesthesia. 2018;73(6):738-745. doi: 10.1111/anae.14181. 

Presented by: Dr Megan Burton


This study looks at monitoring activity in patients before major surgery with the aim to identify patients at high-risk of peri-operative morbidity and mortality.

Cardio-pulmonary exercise testing (CPET) is used to access cardiac and pulmonary reserve and identify patients at high-risk of mobility and mortality. However CPET is restricted to particular hospitals and select patients.

This study’s aim is to try and use home-worn accelerometers as a cheaper and more accessible way of measuring pre-operative physical activity to predict adverse outcome in patient undergoing major surgery compared with cardio-pulmonary exercise testing.

Design & Setting

A UK based study comparing methods of measurement of pre- operative cardio-pulmonary fitness. Compares variables measured using a waterproof chest accelerometer which is fitted to the patient at their CPET and is worn continuously at home fore a 72 hour period.

These measurements are compared with Pre-operative cardio-pulmonary testing variables:

  • Peak power
  • Peak Oxygen consumption
  • Anaerobic threshold
  • Ventilatory equivalents for O2 and CO2
  • Predicted values for peak oxygen consumption.

Subjects were also asked to complete two self-reporting questionnaires to estimate physical activity and functional reserve:

  1. The Duke activity status index (DASI)
  2. The SF-12.

The accelerometer measures are compared to PRET measurements using persons correlation coefficient.



Recruited participants were recruited from patients attending the pre- operative cardiopulmonary exercise clinic.

  • 50 recruited majority of whom were male (42).
  • 48 completed the 72 hours wearing the accelerometer.
  • Mean age 70,
  • Majority ASA 2’s and 3’s.
  • Mean DASI score = 37.6 and mean SF-12 = 44.4.

DASI is a subjective estimate of functional capacity. Maximum score is 58.2, and a score of 37.6 suggests pretty high functionality. i.e. able to walk up a hill, do yard work, climb a flight of stairs.

The SF-12 is a subjective measure of health and well being. the mean score for the US general population is 50.


Participates had a waterproof accelerometer strapped to their chest for 72 hours. Accelerometer measures were collected and compared to cardio- pulmonary testing variables, which were collected from a standard CPET involving incremental exercise on a stationary bike whilst attached to ECG, pulse oximetry and gas analysis.This provided information on a number of variables: Peak power, Peak Oxygen consumption, anaerobic threshold, ventilator equivalents for O2 and CO2 and predicted values for peak oxygen consumption. They also asked participants to complete self reporting questionnaires: The Duke activity Status index and the SF-12 to look at baseline exercise capacity.


Primary outcome

Explore the associations between accelerometer variables measured at home with variables measured during cardiopulmonary exercise testing in patients scheduled for major surgery.

Secondary Outcome

Assess the acceptability and feasibility of using an accelerometer to determine the pre-operative activity level of patients.

Accelerometer frequency measures were categorised into activity levels (this categorisation has been previously documented and has shown to be 81–93% accurate compared to of videotape analyses)

Activity levels were categorised into Active, Stationary, and Lying, from accelerometer frequencies using an algorithm. These were compared with CPET variables: Peak power, Peak Oxygen consumption, anaerobic threshold, ventilator equivalents for O2 and CO2 and predicted values for peak oxygen consumption. These were compared using persons correlation coefficient to give an r value.

NB. Persons correlation coefficient is a measure of the linear correlation between two variables X and Y. It has a value between +1 and −1, where 1 is total positive linear correlation, 0 is no linear correlation, and −1 is total negative linear correlation.


Accelerometer values:

  1. Lying 42.6% (12.6%) – No correlation with CPET variables
  2. stationary 53.0% (11.9%) – No correlation with CPET variables
  3. Active 4.7% (3.1%) – linearly correlated with: peak power, Peak O2 consumption, and anaerobic threshold ( r=0.5-0.7). Some correlation with ventilator equivalents for O2 and CO2 (r=<0.5). And no correlation with Predicted values for peak O2 consumption.


Pre-operative accelerometry is feasible and the study did find an association with fitness measured by cardiopulmonary exercise tests. But the accelerometer measurements only correlated when the participants were active, when they were inactive the measure’s did not correlate with CPET variables. There was a large amount of inactivity recorded; 95% were inactive for 24hrs of the day and sedentary for 99% of the time.

Overall the main conclusion they were able to produce was that pre-operative accelerometer is feasible and that its the duration of moderate intensity activity is likely to be the most useful measure of fitness.



  1. Adequately powered study: The study calculated that a cohort of 47 participants would have 80% power to identify a Pearson’s correlation coefficient ≥ 0.4 at p ≤ 0.05 between an activity variable and either anaerobic threshold or peak oxygen consumption. They achieved 48 participants.
  2. The study was able to show correlation when active between the accelerometer readings and the CPET measurements with an r value of 0.5-0.7 which was the primary aim of the study.
  3. The secondary aim was the access the acceptability and feasibility of using an accelerometer to determine the pre-operative activity level of patients. Patients apparently reported that wearing an accelerometer at home for three consecutive days was acceptable (although 1/4 didn’t complete the 72hr period!).


  1. Correlation coefficients are not robust so the r value can be misleading and although is widespread method in the analysis of comparing medical testing methods it is generally not considered a very good way.
  2. Chest accelerometers may not pick up activity where the chest is still but the legs are active- such as cycling. Considering patients underwent their CPET on a bicycle it seems unfortunate to miss potential data in this way.
  3. 72 hours is a snap shot of activity and doesn’t predict the potential for activity. Many did not complete the full 72 hour period so may miss some activity. And may not be a good predictor of cardio-respiratory reserve.
  4. One-quarter of the participants did not wear the accelerometer for 72 hrs, which might have limited how representative the results were for the total cohort.


I’m not convinced the chest accelerometer has the potential to replace CPET in predicting function reserve but perhaps there is a place for their use in the implementation in preoperative cardio-pulmonary exercise interventions. The study suggests that longitudinal methods of daily activity will be used to supplement formal cardiorespiratory fitness assessments but I don’t think theres enough evidence from this study to use it as a productive measure.

Potential for impact?

It may be useful as an activity measuring method in future research for the implementation of pre-operative exercise interventions.