The health surveys
A study was designed to assess the social and health benefits to tenants resulting from the housing improvements. In the absence of appropriate comparator groups within the area, it was decided to follow the tenants throughout the upgrade programme with a series of health surveys. The design consisted of five repeated cross-sectional health surveys conducted over a seven-year period from 2009 to 2016. The sampling frame of the survey were social housing tenants of the council. The social housing stock of the council consisted of approximately 9200 properties at the time of the study, of which about 400 houses were of non-traditional build (e.g. steel-framed or concrete constructions). All addresses were approached for either a postal survey or face-to-face interviews. Any adult tenant currently living in the council-owned properties was eligible for inclusion. Responses from non-traditional properties were excluded from the analyses, to ensure that the results of the study are generalisable to other programmes upgrading standard social housing in the UK.
The first wave of data collection took place in early 2009 as part of a pilot study. Following the successful delivery of the pilot study, four additional health surveys were conducted in 2011, 2012, 2014 and 2016.
7 All five surveys that are part of the study were conducted in the same period within the heating season. The fifth and final survey was conducted at the end of the programme, when most of the improvements were completed. Data were collected using postal questionnaire and face-to-face interviews by two Wales-based market research companies with established networks of trained interviewers. The project received ethical approval from the School Research Ethics Committee of the Welsh School of Architecture.
In the first survey, conducted in 2009, tenants of properties where work had not started yet were contacted for face-to-face interviews. All other tenants were approached by postal questionnaire. All tenants who took part in the first survey were contacted to be interviewed face-to-face for the second survey in 2011. The remaining tenants were contacted by postal questionnaire. Respondents who had not consented to be re-contacted were excluded from the study. This sampling strategy was repeated in the remaining surveys to ensure that the samples were comparable across the different years. The response rates were generally higher for the face-to-face interviews than for the postal questionnaires.
8 In all five surveys, a prize draw was offered as an incentive for residents to complete and return their postal questionnaires. The number of households contacted decreased over the seven-year period of the study. This was mostly due to the exclusion of tenants who did not want to be contacted again.
The survey was designed to cover the topics of (1) housing suitability, satisfaction, and quality, (2) thermal comfort and household finances, and (3) mental, respiratory and general health. The three topics are used as a structure for this paper.
Data from the five health surveys were combined to form a single dataset, and subsequently linked to intervention data held by the council (see
The intervention dataset section). The data were linked using a unique property reference number, which was included in the surveys and in the intervention dataset. The reference number enabled the properties to be matched with deprivation data from the Welsh Index of Multiple Deprivation. After removal of multiple responses from the same household and of responses from tenants living in non-traditional housing, the overall dataset consisted of 10,009 individual responses. Respondent characteristics of the five surveys are provided in Table
2.
Table 2
Socio-demographic characteristics of the respondents to the five surveys (%)
Male | 31 | 35 | 36 | 35 | 35 |
Age under 36 | 15 | 15 | 15 | 15 | 12 |
Age 36–45 | 12 | 11 | 10 | 10 | 10 |
Age 46–54 | 11 | 11 | 12 | 11 | 12 |
Age 55–64 | 17 | 16 | 17 | 16 | 16 |
Age 65+ | 41 | 44 | 45 | 48 | 48 |
No qualification | 39 | 56 | 58 | 54 | 49 |
Housing benefit recipient | 61 | 68 | 70 | 70 | 68 |
WIMD quartile 1 | 19 | 24 | 20 | 21 | 18 |
WIMD quartile 2 | 37 | 36 | 36 | 37 | 38 |
WIMD quartile 3 | 32 | 29 | 31 | 30 | 31 |
WIMD quartile 4 | 13 | 12 | 13 | 13 | 14 |
Retired | 35 | 43 | 45 | 47 | 43 |
Not working other | 36 | 36 | 37 | 33 | 34 |
Current smoker | 34 | 33 | 34 | 30 | 29 |
Past smoker | 29 | 29 | 24 | 29 | 29 |
Face to face interview | 15 | 44 | 50 | 52 | 53 |
Postal questionnaire | 85 | 56 | 50 | 48 | 47 |
The intervention dataset
Records on progress of the upgrade programme were used to determine changes in the status of
eight housing intervention measures, i.e. (1) new windows and doors, (2) boilers, (3) kitchens, (4) bathrooms, (5) electrics, (6) loft insulation, (7) cavity-wall insulation, and (8) external wall insulation. Measures that were part of the gardens and estates work package were not included because this part of the programme was scaled down at an early stage. For each survey, properties were categorised according to whether an intervention had taken place or not, for each of the listed measures. Table
3 shows the number and percentage of completed measures in the five surveys.
Table 3
Number and percentage of completed individual housing intervention measures in the five surveys
Windows and doors | 1083 (52%) | 2138 (96%) | 1996 (99%) | 1983 (100%) | 1709 (100%) |
Boilers | 375 (18%) | 878 (40%) | 985 (49%) | 1455 (73%) | 1435 (84%) |
Kitchens | 248 (12%) | 762 (34%) | 864 (43%) | 1466 (74%) | 1551 (91%) |
Bathrooms | 327 (16%) | 872 (39%) | 945 (47%) | 1543 (78%) | 1581 (93%) |
Electrics | 318 (15%) | 903 (41%) | 968 (48%) | 1549 (78%) | 1628 (96%) |
Loft insulation | 1129 (55%) | 1754 (79%) | 1611 (80%) | 1608 (81%) | 1381 (81%) |
Cavity wall insulation | 1533 (74%) | 1706 (77%) | 1554 (77%) | 1581 (79%) | 1385 (81%) |
External wall insulation | 110 (5%) | 154 (7%) | 184 (9%) | 200 (10%) | 222 (13%) |
Number of measures per property (SD) | 2.46 (1.51) | 4.13 (1.83) | 4.52 (1.88) | 5.72 (1.63) | 6.37 (1.02) |
Total spend: £k/property (SD) | 5.45 (5.29) | 10.56 (6.70) | 12.09 (6.91) | 16.28 (5.95) | 18.67 (4.37) |
The number of improvements installed in each property varied according to the condition of the property at the start of the study. Some properties were already compliant for some or all of the measures at the time of the first survey in 2009. Where properties had received measures prior to the first survey, they were categorised as having received the intervention. The timing of the work programme was not related to residents’ needs in terms of health, housing condition, or other personal characteristics, and can therefore be assumed effectively random. Where properties had received measures prior to the first survey round, they were still categorised as receiving an intervention and the date was recorded for the installation of each measure.
The intervention data were further used to calculate a number of measures variable. This variable indicates how many different intervention measures (out of the eight housing measures of interest) had been installed within each property at the time the five surveys were conducted. The maximum number of possible measures in any one property was seven (a property could only receive external wall or cavity-wall insulation).
A
total spend variable was created reflecting the average total investment of the programme. The variable was calculated for each individual property at the time of the five surveys and expressed in thousand pounds per property (£k/property), using cost data provided by the council. The
total spend variable is closely related to the
number of measures variables, but weighted according to the cost of the individual measures.
9 This is based on the assumptions that the cost of the measure is indicative of its size and therefore potential impact on the outcome measures.
Outcome measures
The study focused on nine outcome measures relating to the topics of housing suitability, satisfaction, and quality; thermal comfort and household finances; and mental, respiratory and general health.
Three measures were used to indicate housing suitability, satisfaction, and quality, respectively.
Suitability of housing was measured by asking the tenants “Do you feel that your home is suitable for the needs of you and your family?” Response options were “yes” (1) or “no” (0).
Satisfaction with the state of repair of the home was measured by asking “In general, how satisfied or dissatisfied are you with the current state of repair of your home?” The 5-point answering scale ranged from “very dissatisfied” (1) to “very satisfied” (5), with “neither satisfied nor dissatisfied” as the scale midpoint (3).
A building problems variable was constructed as an indicator of housing quality. Respondents were asked whether they currently experience any of the following problems in their home: draught, condensation, damp on walls and/or floors, mould, lack of adequate heating, rot in windows or doorframes, and leaking roof. This variable was dichotomised to indicate whether someone had experienced any building problems (1) or not (0).
Three measures were used to indicate thermal comfort and household finances.
Thermal comfort was measured by asking tenants the extent to which they were satisfied or dissatisfied with the temperature in (1) the main living room during the day, (2) the main living room in the evening, and (3) the bedroom at night on a typical winter’s day. Scale analyses showed that the reliability of the scale is consistently high across the five surveys (Cronbach’s α > 0.90). The three items were therefore combined into a single thermal comfort scale ranging from 1 (strongly dissatisfied) to 5 (strongly satisfied).
Household finances was measured using two items. The Costs of living in home was determined by asking tenants “How easy or difficult is it for you to meet the costs of living in this home?” A 5-point answering scale ranged from “very easy” (1) to “very difficult” (5), with “neither easy nor difficult” as the scale midpoint (3). The Difficulties paying utility bills variable reflects responses to the question: “In the past 12 months, have you had difficulties paying scheduled utility bills, such as electricity, water, gas?” The response options were “yes” (1) and “no” (0).
Three mental, respiratory and general health outcomes were used in the study.
A mental health scale was created by averaging the answers to five questions relating to how much of the time in the previous 4 weeks respondents have been a very nervous person (reversed), have felt so down in the dumps that nothing could cheer them up (reversed), have felt calm and peaceful, have felt downhearted and blue (reversed), and have been a happy person. Scale analyses showed that the five items could be combined to create an internally consistent scale (Cronbach’s alpha ≥0.85 across the five surveys). The frequency response options ranged from “none of the time” to “all of the time”, which were subsequently weighted to form a scale ranging from 0 to 100.
Respiratory health was assessed by asking respondents to report whether they had experienced any of eleven listed respiratory symptoms in the past month. The symptoms included coughing, bringing up phlegm, shortness of breath, wheezing, chest tightness, runny nose, blocked nose, sinus swelling, sneezing, sore throat or fever. The responses were dichotomised, reflecting whether respondents had experienced any respiratory symptoms (1) or not (0).
The
general health variable reflects responses to the question “In general, how would you say your health is?” Respondents could use the following categories: excellent, very good, good, fair, poor. This variable was dichotomised to contrast excellent, very good and good (1) with fair and poor (0) health. The self-rated health variable has been validated as a measure of general health and mortality [
34‐
36].