Results of a 15-year systematic survey of commensal rodents in English dwellings

Population trends for commensal rodents are the subject of interest and speculation but accurate data are rarely available. Here we report data from a 15-year systematic survey of rats and mice in English dwellings and present national-level estimates of prevalence for 1996–2010. We found evidence for peaks in prevalence of mice inside and rats around dwellings in 2002 and 2008. Models containing twelve variables relating to the dwelling and local area explained some but not all of the variation in prevalence. Older dwellings, those in rural areas, those with litter, drainage faults and pets or other animals outdoors tended to have higher levels of rodent prevalence. Regional differences were found but there were no seasonal trends apart from lower prevalence of mice during summer. Rodent prevalence was generally higher in rented (compared to owner-occupied) dwellings, although apparently not due to reduced tendency to carry out rodent control. The percentage of households having taken some form of action against active rodent problems varied according to prevalence at the national level, and therefore appropriate data on number of rodent control treatments carried out each year could likely act as a useful index of household rodent prevalence.

 Unmade/no roadtypically a track to a farmhouse or remote cottage.

Drainage system faults;
Survey data for presence or absence of faults in underground drainage including blockages or other faults relevant to Health and Safety risks were aggregated into a single yes/no response.

Tenure type;
Tenures relates to the ownership of the dwelling.     Other -converted flats (buildings converted into individual flats which have been defined as separate dwellings and non-residential plus flat where residents must pass through nonresidential accommodation to gain access to their living accommodation.

Date of construction;
Date of original construction. If a property had a large later extension or had been partially rebuilt, age of the oldest part was recorded even if it accounted for less than half of the area of dwelling.

Region;
Location of dwelling by Government Office Region (GOR).
Variables relating to the LOCAL AREA.
Surveyors were asked to define the 'area around the dwelling of which the dwelling seems to be a part' and estimate the number of dwellings in that area. This Local Area was likely to be, but was not necessarily, defined by physical boundaries such as roads, railway lines, canals, etc. The survey dwelling was not necessary at the centre of the area; surveyors were asked to define an area of manageable size so that they could clearly define the boundaries of the local area and visually inspect the whole area on foot before proceeding to complete the following questions:  Commercial City/Town Centrethis is the area that would constitute part/all of the centre of a city or town. Areas do not have to be run down to be coded as city or town centre. It is likely that these areas will have a high percentage of commercial properties such as shops and businesses.
 Urbanthis is the area around the core of towns and cities, and also older urban areas which have been swallowed up by a metropolis. Areas would be largely but not exclusively residential.
 Suburban residentialthis is the outer area of towns or cities, and would include large, planned housing estates on the outskirts of towns or larger areas of older residential stock.
 Rural residentialthese can be free standing residential areas or suburban areas of villages, often meeting the housing needs of people who work in nearby towns and cities.
 Village centrethese are traditional English villages or the old heart of villages which have been suburbanised.
 Ruralthese areas are predominantly rural e.g. agricultural with isolated dwellings or small hamlets.

Problems in area;
Depending on the year of the survey, up to 16 categories of problems in the local area, such as litter, neglected buildings, scruffy gardens, vacant buildings and vandalism, were scored by the surveyors on a scale of 1 (no problems) to 5 (major problems). The mean score from 14 of these categories (x̄) was used to allocate dwellings to one of four categories; no problems (x̄ = 1), slight problems (1 < x̄ < 1.5), moderate problems (1.5 < x̄ < 2.0) or substantial problems (x̄  2.0). This variable was treated as a factor in the models; the 14 categories (selected because they were recorded for all survey years) were vacant sites, intrusive industry, non-conforming uses, vacant boarded-up buildings, ambient air quality, heavy traffic, intrusion from motorways and arterial roads, noise from railways or aircraft, nuisance from on-street parking, litter and rubbish, scruffy gardens, vandalism, graffiti, and scruffy or neglected buildings.

Rodent control;
Where the householder reported a current problem the surveyor asked to see evidence and asked two additional questions;  Has anyone treated the rats/mice problem?
 Is anything currently being done to stop or control the rats/mice problem?
If the answer to the first question was yes, further details were recorded on how the problem was treated and by whom.  Supplementary Figure S7: Estimated smoothed function, using thin plate regression splines, for year of survey on the percentage of dwellings with a current rodent problem (and where information on rodent control arrangements was recorded) having a) done something about the problem and b) taking action at the time of the survey.The solid line is the smoother and the dotted lines are the 95% confidence bands. The y-axis represents the value taken by the centered smoother. It is the contribution (at a value of the covariate) made to the fitted value for that smooth function.

Supplementary
The model selection process suggested the use of generalized additive models (GAM) fitted with binomial errors and logit link function to model the effects of year of survey as a smoothed function on the percentage of dwellings with a current rodent problem (and where information on rodent control arrangements was recorded) having a) done something about the problem (df = 8.34, deviance, 30.58, p<0.001) and b) taking action at the time of the survey (df = 8.49, deviance, 79.42, p<0.001), with the estimated smoothers showing pronounced non-linear effects.