Carbon monoxide exposure inside UK road vehicles: a pilot study
Data files
Jul 21, 2025 version files 6.95 MB
-
README.md
3.25 KB
-
Stats_bundle_updated.zip
6.94 MB
Abstract
Objective
To test the following hypotheses:
(i) CO is present inside the passenger cabins of road vehicles driven by members of the general public in the UK;
(ii) in-cabin CO is due, at least in part, to internal exhaust leakage; and
(iii) the use of handheld dataloggers by laypeople can generate data relevant to the above two objectives.
Design
Pilot study.
Setting
Two centres: privately-owned cars in Chesham and Amersham, and cars used by engineers at Southern Gas Networks in Epsom.
Participants
28 participants entered the study, through online local recruitment (first centre, n=21) and through line management (second centre, n=7), driving 33 vehicles. The study excluded vehicles carrying smokers.
Primary outcome measure
Parts per million CO, logged continuously during journeys. Mean journey per cent ppm CO was calculated and peak journey ppm CO noted.
Methods
Measurement of CO using mobile-compatible dataloggers.
Results
33 vehicles returned 230 journey logs in all. 20 (61%) of cars logged non-zero CO. Mean all-journey average was 6.629ppm CO. 10 journeys measured ≥10ppm CO at least once. Peak single-journey average CO was 192.174 ppm. Some patterns of CO ingress were suggestive of internal exhaust leakage.
Conclusions
This is the first public-engagement UK-based study of CO levels within vehicles. It shows that in-cabin average CO levels are non-zero overall, and in some cases markedly raised.
Chronic low-level CO exposure has a range of harmful effects. In addition to causing hypoxic stress, it contributes to cardiovascular disease, generation of reactive oxygen species, and demyelination of white matter. Pregnant women, the unborn and children are especially vulnerable to its effects, which include gestational-specific harms.
Adding in-car air quality measurement to the MOT would benefit public health, alerting vehicle owners, who may be asymptomatic, to raised in-vehicle exhaust levels. Routine MOT air quality testing would have the additional benefit of capturing wider data on the problem, as would larger studies of similar design.
Data for the www.airsafe.london project, a pilot study funded for the Carbon Monoxide Research Trust
This dataset holds a record of each datalog recorded by testers measuring in-cabin vehicle COppm. Most datalogs appear in table and graph form, although in some instances, logs appear in graph format only.
Description of the data and file structure
Each datalog is saved as a separate file.
Formatting of .csv files
The .csv files are formatted as follows:
The first eight lines are human-readable metadata (including the recording date and sampling frequency) and summary information. Line 10 contains the headers of the main data; line 11 onwards contains the data itself.
The data is of the form
,x,y,u
The datalogging software, which ran on a range of mobile phone operating systems, did not generate a 'u' column in all tables: however, all values show measurements made in units of ppm.
where x is the sample number, y is the recorded value, and u (where present) is the unit (always ppm meaning parts per million).
File naming convention
The file naming convention uses the following format:
[Anonymised tester number]_[log number]_[table/graph].
Tables and graphs carrying the same input data have the same tester and log numbers.
Testers 2 and 6 returned graphs but not tables. Where datalogs returned by these testers captured non-zero levels of CO, an accompanying .csv file (with duplicate .xlsx file) presents the graph data in table form.
Where datalogs were generated by means of a secondary datalogger as a cross-check, the additional field [Easylog] appears after the first three fields set out above. Easylog files do not carry a corresponding graph. Data appears in .csv form and is duplicated in .txt form.
In logs 14_8 and 17_1, levels of CO captured exceeded the 999ppm maximum calibration of the datalogger (see section 2.8 of the write-up). We have accordingly capped these values off at 999ppm, adding the additional field [adjusted] after the first three fields.
Datalogs from testers 3, 4, and 16 subdivide as follows (vehicle numbers refer to those listed at Table 1 in the write-up):
Tester 3:
Log 3_1 shows values measured inside vehicle no. 4
Log 3_2 shows values measured inside vehicle no. 3
Log 3_3 shows values measured inside vehicle no.5
Tester 4:
Logs 4_1 to 4_4 (inclusive) show values measured inside vehicle no. 7.
Logs 4_5 to 4_27 (inclusive) show values measured inside vehicle no. 6.
Tester 16:
Log 16_1 shows values measured inside vehicle no. 19.
Log 16_2 shows values measured inside vehicle no. 20.
Image files
The software used also exported image files of the data as graphs. They are included as a convenience. Some do not have file extensions, but they are .png files and can be opened by any image viewer.
Sharing/Access information
Datalogs of note feature in presentations given by the study author as part of the Carbon Monoxide Research Trust 2022 Winter Lecture Series and as part of the 2024 NCOAA conference.
They are also reproduced in the write-up to this study, which is to be resubmitted for publication during summer 2024.
Testers measured in-cabin COppm, using the Kane COA1 Carbon Monoxide Detector Adapter. Although the maximum range of the sensor was 999 COppm, the app supporting the datalogger generated some graphs with readings in excess of this level. For calculation purposes, values have been capped at 999ppm.
Data is left in arithmetic form, with right-skewing unaltered.
Data analysis was performed with custom scripts written in Lua.
Readings from one datalog (see Supplemental Material) were set aside due to likely artefactual interference on the sensor.