- Email: [email protected]
Annoyance and health symptoms and their influencing factors: A population-based air pollution intervention study
Public Health 123 (2009) 339–345
Contents lists available at ScienceDirect
Public Health journal homepage: www.elsevierhealth.com/journals/pubh
Original Research
Annoyance and health symptoms and their influencing factors: A population-based air pollution intervention study T. Stenlund a, b, E. Lide´n c, K. Andersson c, J. Garvill a, S. Nordin a, c, * a
Department of Psychology, Umeå University, Sweden Department of Educational Measurement, Umeå University, Sweden c ¨ rebro University Hospital, Sweden Department of Occupational and Environmental Medicine, O b
a r t i c l e i n f o
s u m m a r y
Article history: Received 25 June 2008 Received in revised form 13 November 2008 Accepted 16 December 2008 Available online 3 April 2009
Objectives: Interventions for reducing air pollution are important means for improving public health. The role of psychological factors in understanding annoyance and health symptoms due to air pollution is limited and further investigation is required. This study aimed to investigate the effects of an intervention to reduce air pollution (predominantly dust and soot) with respect to perceived pollution, risk perception, annoyance and health symptoms. Another objective was to test a model that describes interrelations between air pollution, perceived pollution, health risk perception, annoyance and health symptoms.
Keywords: Epidemiology Dust Soot Odour Perception Risk
Study design: An interventional, population-based questionnaire study. Methods: Surveys were performed before (pre-test) and after (post-test) closure of a sinter plant. Instead, pellets were shipped to the community’s harbour for steel production. Individuals in the community aged 18–75 years were selected at random for participation in the pre-test (n ¼ 738; 74% of the sample participated) and post-test (n ¼ 684; 68% of the sample participated). The two samples were representative of the populations at the two points in time, and thus not identical. Results: After the sinter plant was closed, the environment was perceived as being less dusty, the residents were more positive in their risk perception, and they reported less annoyance due to dust, soot and odorous substances. No difference was found for health symptoms between the pre-test and the posttest. Based on path analyses, a model is proposed of inter-relations between air pollution, perceived pollution, health risk perception, annoyance and health symptoms. Conclusion: The intervention was successful with respect to perceived dust and soot pollution; to annoyance attributed to dust, soot and odorous substances; and to risk perception. The path analyses suggest that perceived pollution and health risk perception play important roles in understanding and predicting environmentally induced annoyance and health symptoms. Ó 2008 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
Introduction Effects of air pollution on mortality due to cancer, cardiovascular disease and pulmonary disease are well documented,1 and the societal costs attributed to air-pollution-related health hazards are considerable.2 As a consequence, public concern about the effects of air pollution has increased over the past decades.3 Two common effects of air pollution are annoyance4 and health-related symptoms.5,6 The concept of annoyance is complex and can be considered a mixture of perception, emotions and attitudes.7 The healthrelated symptoms vary depending on the category of pollutant, such * Corresponding author. Department of Psychology, Umeå University, SE-901 87 Umeå, Sweden. Tel.: þ46 90 7866006; fax: þ46 90 7866695. E-mail address: [email protected] (S. Nordin).
as gaseous pollutants and particulate matter. Common gaseous urban pollutants are sulphur dioxide, nitrogen dioxide and ozone, which predominantly cause pulmonary symptoms,8,9 but also neurasthenic symptoms10,11 and eye, nose and throat irritation.12 Particulate matter typically consists of dust and soot, causing upper respiratory and pulmonary symptoms.13 The respiratory symptoms from urban air pollution resemble bronchitis14 and asthma; the latter predominantly among asthmatics.15 With regards to both annoyance and symptoms, perceived pollution and risk perception are likely to play important roles. Perception of air pollution is predominantly based on visual and chemosensory cues, and has been shown to be a good indicator of air pollution16 and to mediate between environmental exposure and health.17 A commonly used definition of risk perception is that it involves people’s beliefs, attitudes, judgements and feelings.18 In
0033-3506/$ – see front matter Ó 2008 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.puhe.2008.12.021
340
T. Stenlund et al. / Public Health 123 (2009) 339–345
this respect, feelings may include worries of compromised health attributed to the environmental exposure. A survey was conducted in 1989 among residents in the community of Oxelo¨sund, which is located on the east coast of Sweden, approximately 110 km south of Stockholm. Oxelo¨sund is dominated by steel and manufacturing industry. The survey showed that air pollution, predominantly dust and soot, was considered to be a major environmental problem.19 The pollution was related to the steel industry in this community, and it was estimated that approximately 600 tons of dust and soot was released in the air each year. The dominant substances were sulphur dioxide, nitrogen dioxide and ozone. Benzene was another salient pollutant. In addition to evoking odour sensations, these substances activate the chemosomatosensory system at relatively weak concentrations, leading to eye irritation and a pungent nasal sensation. As a consequence of the results from the survey in 1989, and in order to reduce air contamination, the steel industry in Oxelo¨sund closed down its sinter plant in Summer 1995. Instead, pellets were shipped to Oxelo¨sund Harbour for steel production. As a result of this change and other industrial ventures, it was hypothesized that the residents in Oxelo¨sund would: (1) perceive the environment as being less dusty, sooty and odorous; (2) become more positive in their risk perception (beliefs, attitudes and judgements); (3) report less annoyance due to air pollution; and (4) report fewer symptoms. An intervention study was undertaken to investigate this hypothesis. A first survey (pre-test) was conducted in Spring 1995 and a second survey (post-test) was conducted 3 years later in Spring 1998. This study aimed to investigate the effects of the intervention (i.e. closure of the sinter plant) by comparing the pre- and post-test surveys with respect to perceived pollution, risk perception, annoyance and health symptoms. There is a lack of investigations that, by means of observing the same individuals, have studied inter-relations between variables that are expected to influence annoyance and health symptoms. A second objective of this study was therefore to test and further develop a model that describes inter-relations between air pollution, perceived pollution, health risk perception (worries of compromised health), annoyance and health symptoms. According to previous research, air pollution can be expected to influence perceived pollution, health risk perception, annoyance and health symptoms.5,6,16,20 Perceived pollution can, in turn, be expected to affect health risk perception, annoyance and symptoms.17,20–23 Health risk perception may contribute to annoyance and symptoms,21,24–26 and annoyance can be expected to affect symptoms.16 These inter-relations suggest a model that is presented in Fig. 1, which was tested with path analysis. Questions about health risk perception were included in the post-test survey to enable testing of the model. Methods Out of 8665 inhabitants in Oxelo¨sund in 1995, 1000 individuals aged 18–75 years were selected at random for participation in the
Figure 1. Initial path-analytic model of influences of air pollution, perceived pollution, health risk perception and annoyance on symptoms.
pre-test. A questionnaire was mailed to the participants in May, and duplicate copies were posted to non-respondents after 2 and 13 weeks. This questionnaire was completed by 738 individuals (74%). For the post-test in May 1998, 1016 individuals out of 8111, aged 18– 75 years, were selected at random. The distribution of questionnaires followed that of the pre-test, and 684 (68%) individuals completed this questionnaire. The advantage of the samples being representative of the population at the two time points motivated use of this procedure. Data on demographics and smoking habits are given in Table 1. According to Chi-square analyses, the two samples did not differ significantly with respect to any of the demographic variables, except for the number of children at home, which was higher in the post-test sample. This sample also included a higher proportion of smokers. Seventy-eight individuals participated in both surveys. The samples were drawn by Statistics ¨ rebro. Sweden in O The questions used in this study are given in Tables 2 (perceived pollution), 3 (risk perception), 4 (health risk perception; at posttest only for the path analysis), 5 (annoyance) and 6 (health symptoms; predominantly bronchitis- and asthma-like, and neurasthenic). For the factor ‘air pollution’ in the path analysis, the participants were categorized as being exposed to either a high or low level of pollution, depending on their residential area. High exposure was related to a location relatively close to the steel industry (Swedish post code areas 110, 111, 120, 124, 131 and 210 in
Table 1 Frequency (%) data on demographics and smoking habits for the pre- and post-test samples.
Gender Male Female Marital status Single Married or cohabiting Age (years) 18–29 30–44 45–64 >64 Highest level of education Elementary school High school University Occupation Employed Unemployed Student Retired Time lived in Oxelo¨sund (years) <1 1–5 >5 Time lived in present residence (years) 1–9 10–19 20–29 >29 Number of children at home 0 1–2 >2 Household member employed by SSAB or Oxelo¨sund Harbour Presently or formerly Never Smoking Presently/formerly Never **P<0.01; ***P<0.001; ns, not significant.
Pre-test
Post-test
c2
365 (49.5) 373 (50.5)
343 (50.1) 341 (49.9)
0.10 (ns)
157 (22.5) 541 (77.5)
157 (24.6) 480 (75.4)
0.82 (ns)
93 193 320 132
90 175 292 127
(13.2) (25.6) (42.7) (18.5)
0.28 (ns)
288 (39.9) 316 (43.8) 118 (16.3)
234 (35.9) 296 (45.4) 122 (18.7)
2.82 (ns)
436 58 32 173
400 52 37 162
(61.4) (8.0) (5.7) (24.9)
1.01 (ns)
6 (0.8) 46 (6.2) 686 (93.0)
8 (1.2) 35 (5.2) 636 (93.6)
1.12 (ns)
238 159 73 58
217 155 90 66
(41.1) (29.4) (17.0) (12.5)
3.55 (ns)
508 (68.8) 191 (25.9) 39 (5.3)
342 (50.0) 252 (36.8) 90 (13.2)
64.78***
423 (57.3) 315 (42.7)
407 (59.5) 277 (40.5)
0.74 (ns)
380 (51.5) 358 (48.5)
427 (62.4) 257 (37.6)
17.46***
(12.6) (26.2) (43.4) (17.8)
(62.4) (8.3) (4.6) (24.7)
(45.1) (30.1) (13.8) (11.0)
T. Stenlund et al. / Public Health 123 (2009) 339–345 Table 2 Questions and response frequencies (%) pertaining to perceived pollution for the pre- and post-test. Pre-test Rank the degree of environmental exposure/problems in Oxelo¨sund from 1 (most) to 9 (least) Noise 1–3 4–6 7–9 Dust and fallout 1–3 4–6 7–9 Road traffic 1–3 4–6 7–9 Air pollution other than dust and fallout 1–3 4–6 7–9 Effects on sea coast 1–3 4–6 7–9 Odours 1–3 4–6 7–9 Wood burning 1–3 4–6 7–9 Refuse handling 1–3 4–6 7–9 Other exposures/problems 1–3 4–6 7–9 If you have problems with dust or contamination, where does this occur?a, b Car Laundry House façade Windows Indoors (through ventilator or windows) Garden furniture Boat
Post-test
110 (19.6) 275 (48.9) 177 (31.5)
0.86 (ns)
550 (83.7) 59 (9.0) 48 (7.3)
472 (77.4) 90 (14.8) 48 (7.8)
10.44**
78 (13.3) 299 (51.0) 209 (35.7)
108 (19.4) 251 (45.2) 197 (35.4)
8.33**
375 (61.7) 164 (27.0) 69 (11.3)
366 (62.5) 159 (27.1) 61 (10.4)
0.30 (ns)
358 (59.1) 173 (28.5) 75 (12.4)
384 (65.2) 131 (22.2) 74 (12.6)
6.52 (ns)
216 (34.8) 279 (44.9) 126 (20.3)
183 (31.6) 273 (47.2) 123 (21.2)
1.51 (ns)
65 (11.2) 150 (26.0) 363 (62.8)
99 (18.1) 149 (27.3) 298 (54.6)
12.23***
107 (18.4) 189 (32.5) 286 (49.1)
92 (16.7) 167 (30.3) 293 (53.0)
1.76 (ns)
22 (28.6) 12 (15.6) 43 (55.8)
22 (17.1) 15 (11.6) 92 (71.3)
5.74 (ns)
308 122 276 495 286 442 230
(41.7) (16.5) (37.4) (67.1) (38.8) (59.9) (31.2)
327 132 328 459 264 426 257
(47.8) (19.3) (48.0) (67.1) (38.6) (62.3) (37.6)
Table 3 Questions and response frequencies (%) pertaining to risk perception for the preand post-test.
c2
106 (17.7) 297 (49.5) 197 (32.8)
5.53 (ns) 1.92 (ns) 15.77*** 0.01 (ns) 0.02 (ns) 0.73 (ns) 6.62 (ns)
**P<0.01; ***P<0.001; ns, not significant. a Representing perceived pollution in the path analysis. b Several alternatives could be chosen.
Oxelo¨sund), and low exposure was related to a location relatively distant from this industry (remaining post code areas in Oxelo¨sund). Chi-square analyses were computed to compare frequency distributions across response alternatives for the two test occasions. An alevel of 0.01 was chosen to account for the large number of statistical tests being conducted. Path analysis was used to test the model (Fig. 1), which is a form of structural equation modelling for testing and estimating causal relationships using a combination of statistical data and qualitative causal assumptions. This analysis encourages confirmatory rather than exploratory modelling. It starts with a hypothesis, represents it as a model, operationalizes the constructs, and tests the model. To test the model with the post-test data, the factor ‘air pollution’ comprised the variable ‘air pollution exposure’ that was categorized as high (residential location relatively close to the steel industry) or low (residential location relatively distant from the steel industry). Composite measures were used for perceived
341
Contamination of dust and soot is a major environmental and health problem in Oxelo¨sund Agree Doubtful Disagree Mass media’s description of Oxelo¨sund is too negative Agree Doubtful Disagree Oxelo¨sund is an appropriate area to live and work Agree totally Doubtful Disagree Financial considerations are generally given higher priority than environmental considerations in Oxelo¨sund Agree Doubtful Disagree Certain environmental disturbances must be accepted in order to retain and create new jobs in Oxelo¨sund Agree Doubtful Disagree Contamination of cars and boats is a major environmental problem in Oxelo¨sund Agree Doubtful Disagree Everyone has to waive one’s own convenience for a better urban environment Agree Doubtful Disagree The problem with air pollution in Oxelo¨sund is considerably exaggerated Agree Doubtful Disagree Environmental risks are exaggerated by the environmental organizations Agree Doubtful Disagree
Pre-test
Post-test
c2
637 (86.9) 57 (7.8) 39 (5.3)
487 (72.3) 109 (16.2) 78 (11.5)
46.13***
489 (67.3) 149 (20.5) 89 (12.2)
450 (67.2) 145 (21.6) 75 (11.2)
0.60 (ns)
529 (72.3) 157 (21.4) 46 (6.3)
504 (75.4) 121 (18.1) 43 (6.4)
2.58 (ns)
477 (65.5) 183 (25.1) 68 (9.4)
382 (56.8) 203 (30.2) 87 (12.9)
11.92**
397 (54.2) 165 (22.5) 170 (23.3)
360 (53.3) 170 (25.2) 145 (21.5)
1.47 (ns)
504 (68.7) 125 (17.0) 105 (14.3)
414 (61.3) 169 (25.0) 92 (13.7)
13.71**
539 (73.9) 123 (16.9) 67 (9.2)
451 (67.4) 138 (20.6) 80 (12.0)
7.47 (ns)
280 (38.1) 199 (27.1) 256 (34.8)
352 (52.5) 166 (24.7) 153 (22.8)
33.86***
200 (27.7) 220 (30.5) 302 (41.8)
205 (31.0) 249 (37.6) 208 (31.4)
16.91***
**P<0.01; ***P<0.001; ns, not significant.
pollution (seven items specified in Table 2; Cronbach‘s a ¼ 0.75), health risk perception (10 items specified in Table 4; a ¼ 0.73), annoyance (five items specified in Table 5; a ¼ 0.83) and health symptoms (five items specified in Table 6; a ¼ 0.67) that constituted the number of positive responses for these factors except for annoyance. This latter factor was represented by the mean scale value across items (0 ¼ not at all, 1 ¼ a little, 2 ¼ to some extent, 3 ¼ to a large extent). As the proposed model (Fig. 1) is saturated, i.e. includes all relationships between the variables, the model will fit the data perfectly. The aim was to find a more parsimonious model that would still fit the data. Therefore, the sample was randomly divided into two independent groups of equal size (n ¼ 342). Using one of these groups, the model was first tested and modified and then validated in the second group. Root mean squared error of approximation (RMSEA) and Benler’s comparative fit (CFI) were used as goodness-of-fit indices. An RMSEA value of 0.05 is indicative of a good fit, and a value of 0.08 indicates a reasonable fit.27 Divergence from an RMSEA value of 0.05 can be tested, and a non-significant P-value of close fit
342
T. Stenlund et al. / Public Health 123 (2009) 339–345
Table 4 Questions and response frequencies (%) representing health risk perception in path analysis for the post-test. Have public health conditions in Oxelo¨sund degenerated due to the external environment in the past 10 years? Yes No Do not know Have respiratory problems increased in the Oxelo¨sund population due to air pollution in the past 10 years? Yes No Do not know It is hazardous to health to live in Oxelo¨sund? Yes No Do not know Are the public health conditions in Oxelo¨sund due to the contamination of soot/dust much worse than in other communities of the same size? Yes No Do not know Do your relatives and friends have the opinion that it is hazardous to health to live in Oxelo¨sund? Yes No Do not know Is it the general opinion in Oxelo¨sund that it is hazardous to health to live in this community? Yes No Do not know Is it the general opinion outside Oxelo¨sund that this community is dirty and hazardous to health to live in? Yes No Do not know Are you worried that contamination with soot and dust will give you: Bronchitis? Yes No Do not know Asthma? Yes No Do not know Lung cancer? Yes No Do not know
Table 5 Questions and response frequencies (%) pertaining to annoyance for the pre- and post-test. Pre-test
158 (23.4) 330 (48.9) 187 (27.7)
191 (28.3) 291 (43.0) 194 (28.7) 115 (17.0) 409 (60.6) 151 (22.4)
247 (36.6) 298 (44.2) 129 (19.2)
192 (28.5) 346 (51.3) 136 (20.2)
314 (46.7) 212 (31.5) 147 (21.8)
434 (64.4) 110 (16.3) 130 (19.3)
193 (29.9) 328 (50.9) 124 (19.2) 260 (39.6) 285 (43.4) 111 (17.0) 229 (34.8) 294 (44.7) 135 (20.5)
(PCLOSE) indicates that the RMSEA value is not significantly different from 0.05. A CFI value of 0.95 has been suggested to represent a fairly good fit.28 Results Intervention study Perceived pollution The degree of exposure/problem of dust and fallout was ranked as significantly less in the post-test compared with the pre-test, whereas road traffic and wood burning were ranked as becoming more of a problem. No significant difference was found for noise, air pollution other than dust and fallout, effects on sea coast, odours, refuse handling or other exposures/problems between test occasions. When there was a problem with dust or contamination, house façades were more commonly reported as the affected site in the post-test than in the pre-test. No significant differences of this type over time were found for cars, laundry, windows, indoors, garden furniture or boats (Table 2).
To what extent have you been annoyed by the following exposures in the past 3 months?a Dust/soot from road traffic Not at all A little To some extent To a large extent Dust/soot from industry/workshops/ central boiling plants Not at all A little To some extent To a large extent Odours from industry/workshops Not at all A little To some extent To a large extent Exhausts from buses/trucks Not at all A little To some extent To a large extent Exhausts from cars Not at all A little To some extent To a large extent During the past months, have you experienced any of the following disturbances where you live? Difficulties opening windows due to outside noise Often (every week) Occasionally Never Difficulties falling asleep due to outside noise Often (every week) Occasionally Never Difficulties airing due to poor-quality outside air/odour Often (every week) Occasionally Never Difficulties airing due to outside contamination of dust/soot Often (every week) Occasionally Never Is air pollution so annoying that you have seriously considered moving? Is noise so annoying that you have seriously considered moving?
Post-test
c2
405 206 68 38
(56.5) 389 (59.2) (28.7) 187 (28.5) (9.5) 46 (7.0) (5.3) 35 (5.3)
203 177 123 217
(28.2) (24.6) (17.1) (30.1)
233 224 116 152
(32.1) 304 (48.2) 85.51*** (30.9) 195 (30.9) (16.0) 100 (15.8) (21.0) 32 (5.1)
472 172 46 32
(65.4) 444 (67.3) (23.8) 147 (22.3) (6.4) 46 (7.0) (4.4) 23 (3.5)
1.35 (ns)
397 232 66 32
(54.6) 405 (61.4) (31.9) 191 (28.9) (9.1) 40 (6.1) (4.4) 24 (3.6)
8.02 (ns)
24 (3.3) 22 (3.3) 120 (16.5) 110 (16.5) 584 (80.2) 535 (80.2)
0.01 (ns)
16 (2.2) 9 (1.3) 92 (12.7) 58 (8.7) 619 (85.1) 601 (90.0)
7.62 (ns)
285 163 108 103
3.06 (ns)
(43.2) 53.60*** (24.8) (16.4) (15.6)
45 (6.2) 24 (3.6) 11.09** 227 (31.2) 175 (26.2) 455 (62.6) 470 (70.2)
84 200 444 79
(11.5) 45 (6.8) 10.87** (27.5) 176 (26.5) (61.0) 444 (66.7) (10.7) 40 (5.8) 10.62**
21 (2.8)
26 (3.8)
0.79 (ns)
**P<0.01; ***P<0.001; ns, not significant. a Representing annoyance in the path analysis.
Risk perception In the post-test, the participants agreed to a lesser extent, compared with the pre-test, that contamination of dust and soot is a major environmental and health problem, that financial considerations are usually given higher priority than environmental considerations, and that contamination of cars and boats is a major environmental problem. Furthermore, they agreed to a greater extent in the post-test that the problem with air pollution is considerably exaggerated, and that environmental risks are exaggerated by the environmental organizations. No differences were found between test occasions regarding the mass media’s description, the appropriateness of the community as an area to
T. Stenlund et al. / Public Health 123 (2009) 339–345
343
Table 6 Questions and response frequencies (%) pertaining to health symptoms for the pre- and post-test. For comparison, corresponding percentages are given for neurasthenic and sensory symptoms as reference values for people living in multi-family (n ¼ 10,536) and single-family residences (n ¼ 3729).
Have you had any of the following symptoms in the past 12 months? Chest wheezing Tightness in chest Feeling of dyspnoea when waking up Waken up by coughing attacks Asthma attacks Have you had any of the following symptoms in the past 3 months?b Fatigue Feeling heavy headed Headache Nausea or dizziness Attentional difficulties Eye itching, burning or irritation Nose irritation, congestion or running Hoarseness or dry throat Coughing Face skin dryness or redness
c2
Pre-test
Post-test
130 83 49 173 24
(17.6) (11.2) (6.6) (23.4) (3.3)
129 80 47 188 23
(18.9) (11.7) (6.9) (27.5) (3.4)
0.31 (ns) 0.11 (ns) 0.04 (ns) 2.92 (ns) 0.01 (ns)
538 367 404 169 182 246 303 218 279 111
(72.9) (49.7) (54.7) (22.9) (24.7) (33.3) (41.1) (29.5) (37.8) (15.0)
502 360 488 158 175 245 305 196 276 124
(73.4) (52.6) (71.3) (23.1) (25.6) (35.8) (45.0) (28.7) (40.4) (18.1)
0.05 (ns) 1.14 (ns) 40.72*** 0.02 (ns) 0.13 (ns) 0.91 (ns) 1.94 (ns) 0.13 (ns) 0.95 (ns) 2.47 (ns)
Multi family residencec
Single-family residencec
65.9 45.2 49.8 18.3 21.5 28.4 41.2 31.5 32.9 19.2
62.7 37.4 47.8 12.7 15.3 19.4 34.2 21.6 27.1 11.8
a, b
**P<0.01; ***P<0.001; ns, not significant. a Representing symptoms in the path analysis. b Several alternatives could be chosen. c From Reference 31.
live and work, the acceptance of environmental disturbances to create new jobs, or waiving one’s own convenience for a better urban environment (Table 3). Annoyance Compared with the pre-test, dust or soot from industry, workshops or central boiler plants, and odours from industry or workshops were reported to be less annoying in the post-test, whereas no difference was found regarding dust or soot from road traffic or exhausts from buses, trucks or cars between occasions. Moreover, in the post-test, the participants experienced more frequent difficulties with airing because of poor-quality outside air or odour, as well as outside contamination of dust or soot. The test occasions did not differ with respect to difficulties in opening windows and falling asleep due to outside noise. Finally, the results found that fewer people were seriously considering moving due to air pollution, but not due to noise, in the post-test compared with the pretest (Table 5). Health symptoms In comparing the pre- and post-tests, no significant differences in prevalence were found for any types of pulmonary, neurasthenic or sensory symptoms, with the exception of headache which was reported as more prevalent in the post-test (Table 6). Testing the model Correlations between the variables in the model for each group are given in Table 7. As can be seen, all correlations with two exceptions were positive and significant. Using Amos 5.029 and the maximum likelihood method, the model was estimated for the first group. The analysis showed that with three exceptions, the path coefficients were significant (P < 0.05) and had the expected sign. The exceptions were the paths from air pollution to health risk perception and symptoms, and from perceived pollution to symptoms. The non-significant relationships were excluded and the modified model was re-estimated. In the modified model, all path coefficients were significant (P < 0.05). For the modified model, goodness-of-fit indices indicated a good fit (RMSEA ¼ 0.01, PCLOSE ¼ 0.68, CFI ¼ 0.99). Since the modified model was data driven to some extent, it should be validated on an independent
sample. The modified model was estimated for the second group and again all path coefficients were significant (P < 0.05) and the fit was good (RMSEA < 0.001, PCLOSE ¼ 0.93, CFI ¼ 1.00). The modified model and the validated model are shown in Fig. 2. Discussion Population-based questionnaire surveys were conducted in a Swedish community before and after an intervention that aimed to decrease air pollution (predominantly dust and soot, and also odorous substances). This consisted of closing a sinter plant and shipping pellets to the community’s harbour. As hypothesized, after the intervention, the environment was perceived as being less polluted by dust and soot, the residents were more positive in their risk perception (beliefs, attitudes and judgements), and they reported less annoyance due to dust, soot and odorous substances. In these respects, the intervention can be considered successful. Whereas less annoyance was reported after the intervention, the results did not support the hypothesis of an interventionassociated decrease in health symptoms, with respect to either pulmonary, neurasthenic or sensory symptoms. It should be noted that health symptoms typically become a problem at far higher exposure levels than annoyance.30 However, one can only speculate whether the air pollution, even prior to the intervention, was not prominent enough to evoke considerably more symptoms than is normal in Swedish communities. Reference data of this type for persons living in multi-family (n ¼ 10,536) and single-family residences (n ¼ 3729) are presented in Table 6 for neurasthenic and sensory symptoms.31 Although the symptom prevalences for Oxelo¨sund, both before and after the intervention, are generally Table 7 Correlation coefficients between air pollution (AP), perceived pollution (PP), health risk perception (HRP), annoyance (A) and symptoms (S) for the first/second path analysis.
PP HRP A S
AP
PP
HRP
A
0.29***/0.23*** 0.08 (ns)/0.12* 0.35***/0.45*** 0.01 (ns)/0.12*
0.40***/0.45*** 0.44***/0.39*** 0.14**/0.15**
0.42***/0.43*** 0.25***/0.34***
0.26***/0.26***
*P<0.05; **P<0.01; ***P<0.001; ns, not significant.
344
T. Stenlund et al. / Public Health 123 (2009) 339–345
Figure 2. Final path-analytic model of influences of air pollution, perceived pollution, health risk perception and annoyance on symptoms. Proportions of explained variance (r2) and standardized path coefficients are given for the first/second analysis.
somewhat higher than the reference prevalences, the difference is not large. Thus, this comparison may provide some support for the given speculation. Another possible explanation is that the larger percentage of smokers in the post-test sample contributed to pulmonary symptoms, which would overshadow effects of the intervention on these symptoms. There was also only a weak tendency of perceived pollution of odorous substances becoming weaker/less problematic after the intervention, which opposed the hypothesis. A contributing factor may be that pollution of odorous substances was not perceived as particularly strong/problematic even before the intervention. As would be expected from an intervention that aimed to specifically decrease exposure of dust, soot and odorous substances, neither perceived pollution nor annoyance due to environmental issues other than these were found to change after the intervention. This suggests that the differences regarding dust, soot and odorous substances between the two studied samples are not likely to be explained by differences in response behaviour. The two samples were also similar with respect to all studied demographic variables except for the number of children at home, which was larger in the post-test sample. This may have contributed to increased perceived pollution,22 and possibly to a more negative risk perception in this sample, which would imply that the effect of the intervention on perception of pollution and risk may have been even larger than suggested by the present data. It is worth noting that the neurasthenic symptoms (fatigue, feeling heavy headed, headache, nausea or dizziness, and attention difficulties) were rather high in prevalence, both before and after the intervention, compared with reference values, but this was less so for sensory symptoms (Table 6). It is not yet known to what extent the high prevalence of neurasthenic symptoms, both before and after the intervention, can be related to environmental exposure. The categorization of being exposed to either a high or low level of pollution was based on the assumption that residential areas closer to the steel industry were exposed to higher levels of pollution than areas further away from the industry. Although this may seem reasonable, it calls for certain caution since no chemical exposure data are available. It is possible that the absence of a relationship between air pollution and symptoms can be related to this assumption not being fulfilled. A second objective of this study was to empirically test and further develop a model of air pollution and its psychological and medical consequences. The path analyses suggest that air pollution influences perceived pollution which, in turn, influences health risk perception and annoyance. There is also a direct effect of air pollution on annoyance. Health risk perception has an effect on annoyance and symptoms, whereas annoyance influences symptoms. These findings suggest that perceived pollution and health risk perception do indeed play important roles in understanding
and predicting environmentally induced annoyance and health symptoms. In particular, this is the case for health risk perception which explains a rather large proportion of the variance in outcome variables. However, rather small proportions of the variance in perceived pollution and symptoms are explained by the factors in the model, which calls for certain caution. It is likely that various health issues not included in this study would explain considerable variance in symptoms. It is important to point out that the direction of effects in the proposed model in Fig. 2 calls for caution. Thus, the path analysis does not test the cause–effect direction between factors per se. Instead, the directions in the model are causal assumptions based on theory and/or prior empirics. However, the causal assumptions in this model can be questioned. For example, in contrast to the model, it has been suggested that an association between perceived pollution and health risk perception may be due to health risk perception affecting perceived pollution.23 Hence, it cannot be excluded that perceived pollution and health risk perception mutually affect each other, and that symptoms can affect health risk perception and annoyance. It is also quite possible that if people are annoyed by air pollution, they may take measures to reduce this pollution, although in this particular setting, it may be difficult for individuals to reduce pollution. It is important to point out that the cause–effect directions in the proposed model are the most likely directions. In conclusion, perceived pollution, annoyance attributed to dust, soot and odorous substances, and risk perception were reported to be less pronounced after the intervention compared with prior to the intervention. Path analyses further suggest that perceived pollution and health risk perception play important roles in understanding and predicting environmentally induced annoyance and health symptoms. Acknowledgement The authors wish to thank Svenskt Stål Aktiebolag (SSAB), ¨ rebro, and the EnvironOxelo¨sunds hamn, Statistics Sweden in O ment and Protection Board in Oxelo¨sund for their valuable support, and Inger Fagerlund for excellent assistance. Ethical approval Swedish Data Inspection Board. Funding Grants from the Swedish Environmental Protection Agency and ¨ rebro County Council, Sweden. the Environmental County Fund, O Competing interests None declared. References 1. Theophanides M, Anastassopoulou J, Vasilakos C, Maggos T, Theophanides T. Mortality and pollution in several greek cities. J Environ Sci Health Part A 2007;42:741–6. 2. Pervin T, Gerdtham UG, Lyttkens CH. Societal costs of air pollution-related health hazards: a review of methods and results. Cost Eff Resour Alloc 2008;6:19. 3. Frost K, Frank E, Maibach E. Relative risk in the news media: a quantification of misrepresentation. Am J Public Health 1997;87:842–5. ¨ sterberg K, O ¨ stergren PO. Prevalance of 4. Carlsson F, Karlsson B, Orbaek P, O annoyance attributed to electrical equipment and smells in a Swedish population, and relationship with subjective health and daily function. Public Health 2005;119:568–77. 5. Pope CA. What do epidemiologic findings tell us about health effects of environmental aerosols? J Aerosol Med 2000;13:335–54. 6. Leikauf GD. Hazardous air pollutants and asthma. Environ Health Perspect 2002;110(Suppl. 4):505–26.
T. Stenlund et al. / Public Health 123 (2009) 339–345 7. Berglund B, Berglund U, Lindvall T. Measurement and control of annoyance. In: Koelga HS, editor. Environmental annoyance: characterization, measurement and control. Amsterdam: Elsevier Science; 1987. p. 29–43. 8. Neher JO, Koenig JQ. Health effects of outdoor air pollution. Am Fam Physician 1994;49:1397–404. 9. Chen TM, Gokhale J, Shofer S, Kuschner WG. Outdoor air pollution: nitrogen dioxide, sulphur dioxide and carbon monoxide health effects. Am J Med Sci 2007;333:249–56. 10. Paz C. Some consequences of ozone exposure on health. Arch Med Res 1997;28:163–70. 11. Nattero G, Enrico A. Outdoor pollution and headache. Headache 1996;36:243–5. 12. Strøm-Tejsen P, Weschler CJ, Wargocki P, Mysko´w D, Zarzycka J. The influence of ozone on self-evaluation of symptoms in a simulated aircraft cabin. J Expo Sci Environ Epidemiol 2008;18:272–81. 13. Kampa M, Castanas E. Human health effects of air pollution. Environ Pollut 2008;151:362–7. 14. Sunyer J, Jarvis D, Gotschi T, Garcia-Esteban R, Jacquemin B, Aguilera I, et al. Chronic bronchitis and urban air pollution in an international study. Occup Environ Med 2006;63:836–43. 15. Maynard RL. Asthma and urban air pollution. Clin Exp Allergy 2001;31:518–20. 16. Oglesby L, Kunzli N, Monn C, Schindler C, Ackermann-Liebrich U, Leuenberger P. Validity of annoyance scores for estimation of long term air pollution exposure in epidemiologic studies: the Swiss Study on Air Pollution and Lung Diseases in Adults (SAPALDIA). Am J Epidemiol 2000;152:75–83. 17. Elliott SJ, Cole DC, Krueger P, Voorberg N, Wakefield S. The power of perception: health risk attributed to air pollution in an urban industrial neighbourhood. Risk Analysis 1999;19:621–34. 18. Pidgeon NF, Hood C, Jones D, Turner BA, Gibson R. Risk perception. In: Risk analysis, perception and management. London: Royal Society; 1992. p. 89–134. 19. Oxelo¨sund Municipality. Slutrapport PDS 1989-06-01 [Final report from ¨ sunds kommun; 1989. a problem detection study]. Oxelo 20. Bickerstaff K. Risk perception research: socio-cultural perspectives on the public experience of air pollution. Environ Int 2004;30:827–40.
345
21. Petrie KJ, Sivertsen B, Hysing M, Broadbent E, Moss-Morris R, Eriksen HR, et al. Thoroughly modern worries: the relationship of worries about modernity to reported symptoms, health and medical care utilization. J Psychosom Res 2001;51:395–401. 22. Hunter PR, Bickerstaff K, Davies MA. Potential sources of bias in the use of individuals recall of the frequency of exposure to air pollution for use in exposure assessment in epidemiological studies: a cross-sectional survey. Environ Health 2004;31:3. 23. Hunter PR, Davies MA, Hill K, Whittaker M, Sufi F. The prevalence of selfreported symptoms of respiratory disease and community belief about the severity of pollution from various sources. Int J Environ Health Res 2003;13:227–38. 24. Lipscomb JA, Satin KP, Neutra RR. Reported symptom prevalence rates from comparison populations in community-based environmental studies. Arch Environ Health 1992;47:263–9. 25. Williams ID, McCrae IS. Road traffic nuisance in residential and commercial areas. Sci Total Environ 1995;169:75–82. 26. Butani L, O’Connell EJ. Functional respiratory disorders. Ann Allergy Asthma Immunol 1997;79:91–9. 27. Brown MW, Cudeck R. Alternative ways of assessing model fit. In: Bollen KA, Long JS, editors. Testing structural equation models. New Park, CA: Sage; 1993. p. 136–62. 28. Hu L-T, Bentler PM. Cutoff criteria for fit indexes on covariance structure analysis: conventional criteria versus new alternatives. Struct Equat Model 1999;6:1–55. 29. Arbuckle JL, Wothke W. Amos 4.0 user’s guide. Chicago: Small Waters Corporation; 1999. 30. Baird JC, Berglund B, Berglund U, Lindvall T. Symptom patterns as an early warning signal of community health. Environ Int 1990;16:3–9. 31. Andersson K, Norle´n U. Indoor climate and health effects. The indoor climate in the Swedish housing stock. In: Flatheim G, Berg K, Edvardsen K, editors. Indoor air quality in practice. Moisture and cold climate solutions; 1995. p. 23–8. Oslo: Norwegian Society of Chartered Engineers.
Contents lists available at ScienceDirect
Public Health journal homepage: www.elsevierhealth.com/journals/pubh
Original Research
Annoyance and health symptoms and their influencing factors: A population-based air pollution intervention study T. Stenlund a, b, E. Lide´n c, K. Andersson c, J. Garvill a, S. Nordin a, c, * a
Department of Psychology, Umeå University, Sweden Department of Educational Measurement, Umeå University, Sweden c ¨ rebro University Hospital, Sweden Department of Occupational and Environmental Medicine, O b
a r t i c l e i n f o
s u m m a r y
Article history: Received 25 June 2008 Received in revised form 13 November 2008 Accepted 16 December 2008 Available online 3 April 2009
Objectives: Interventions for reducing air pollution are important means for improving public health. The role of psychological factors in understanding annoyance and health symptoms due to air pollution is limited and further investigation is required. This study aimed to investigate the effects of an intervention to reduce air pollution (predominantly dust and soot) with respect to perceived pollution, risk perception, annoyance and health symptoms. Another objective was to test a model that describes interrelations between air pollution, perceived pollution, health risk perception, annoyance and health symptoms.
Keywords: Epidemiology Dust Soot Odour Perception Risk
Study design: An interventional, population-based questionnaire study. Methods: Surveys were performed before (pre-test) and after (post-test) closure of a sinter plant. Instead, pellets were shipped to the community’s harbour for steel production. Individuals in the community aged 18–75 years were selected at random for participation in the pre-test (n ¼ 738; 74% of the sample participated) and post-test (n ¼ 684; 68% of the sample participated). The two samples were representative of the populations at the two points in time, and thus not identical. Results: After the sinter plant was closed, the environment was perceived as being less dusty, the residents were more positive in their risk perception, and they reported less annoyance due to dust, soot and odorous substances. No difference was found for health symptoms between the pre-test and the posttest. Based on path analyses, a model is proposed of inter-relations between air pollution, perceived pollution, health risk perception, annoyance and health symptoms. Conclusion: The intervention was successful with respect to perceived dust and soot pollution; to annoyance attributed to dust, soot and odorous substances; and to risk perception. The path analyses suggest that perceived pollution and health risk perception play important roles in understanding and predicting environmentally induced annoyance and health symptoms. Ó 2008 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
Introduction Effects of air pollution on mortality due to cancer, cardiovascular disease and pulmonary disease are well documented,1 and the societal costs attributed to air-pollution-related health hazards are considerable.2 As a consequence, public concern about the effects of air pollution has increased over the past decades.3 Two common effects of air pollution are annoyance4 and health-related symptoms.5,6 The concept of annoyance is complex and can be considered a mixture of perception, emotions and attitudes.7 The healthrelated symptoms vary depending on the category of pollutant, such * Corresponding author. Department of Psychology, Umeå University, SE-901 87 Umeå, Sweden. Tel.: þ46 90 7866006; fax: þ46 90 7866695. E-mail address: [email protected] (S. Nordin).
as gaseous pollutants and particulate matter. Common gaseous urban pollutants are sulphur dioxide, nitrogen dioxide and ozone, which predominantly cause pulmonary symptoms,8,9 but also neurasthenic symptoms10,11 and eye, nose and throat irritation.12 Particulate matter typically consists of dust and soot, causing upper respiratory and pulmonary symptoms.13 The respiratory symptoms from urban air pollution resemble bronchitis14 and asthma; the latter predominantly among asthmatics.15 With regards to both annoyance and symptoms, perceived pollution and risk perception are likely to play important roles. Perception of air pollution is predominantly based on visual and chemosensory cues, and has been shown to be a good indicator of air pollution16 and to mediate between environmental exposure and health.17 A commonly used definition of risk perception is that it involves people’s beliefs, attitudes, judgements and feelings.18 In
0033-3506/$ – see front matter Ó 2008 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.puhe.2008.12.021
340
T. Stenlund et al. / Public Health 123 (2009) 339–345
this respect, feelings may include worries of compromised health attributed to the environmental exposure. A survey was conducted in 1989 among residents in the community of Oxelo¨sund, which is located on the east coast of Sweden, approximately 110 km south of Stockholm. Oxelo¨sund is dominated by steel and manufacturing industry. The survey showed that air pollution, predominantly dust and soot, was considered to be a major environmental problem.19 The pollution was related to the steel industry in this community, and it was estimated that approximately 600 tons of dust and soot was released in the air each year. The dominant substances were sulphur dioxide, nitrogen dioxide and ozone. Benzene was another salient pollutant. In addition to evoking odour sensations, these substances activate the chemosomatosensory system at relatively weak concentrations, leading to eye irritation and a pungent nasal sensation. As a consequence of the results from the survey in 1989, and in order to reduce air contamination, the steel industry in Oxelo¨sund closed down its sinter plant in Summer 1995. Instead, pellets were shipped to Oxelo¨sund Harbour for steel production. As a result of this change and other industrial ventures, it was hypothesized that the residents in Oxelo¨sund would: (1) perceive the environment as being less dusty, sooty and odorous; (2) become more positive in their risk perception (beliefs, attitudes and judgements); (3) report less annoyance due to air pollution; and (4) report fewer symptoms. An intervention study was undertaken to investigate this hypothesis. A first survey (pre-test) was conducted in Spring 1995 and a second survey (post-test) was conducted 3 years later in Spring 1998. This study aimed to investigate the effects of the intervention (i.e. closure of the sinter plant) by comparing the pre- and post-test surveys with respect to perceived pollution, risk perception, annoyance and health symptoms. There is a lack of investigations that, by means of observing the same individuals, have studied inter-relations between variables that are expected to influence annoyance and health symptoms. A second objective of this study was therefore to test and further develop a model that describes inter-relations between air pollution, perceived pollution, health risk perception (worries of compromised health), annoyance and health symptoms. According to previous research, air pollution can be expected to influence perceived pollution, health risk perception, annoyance and health symptoms.5,6,16,20 Perceived pollution can, in turn, be expected to affect health risk perception, annoyance and symptoms.17,20–23 Health risk perception may contribute to annoyance and symptoms,21,24–26 and annoyance can be expected to affect symptoms.16 These inter-relations suggest a model that is presented in Fig. 1, which was tested with path analysis. Questions about health risk perception were included in the post-test survey to enable testing of the model. Methods Out of 8665 inhabitants in Oxelo¨sund in 1995, 1000 individuals aged 18–75 years were selected at random for participation in the
Figure 1. Initial path-analytic model of influences of air pollution, perceived pollution, health risk perception and annoyance on symptoms.
pre-test. A questionnaire was mailed to the participants in May, and duplicate copies were posted to non-respondents after 2 and 13 weeks. This questionnaire was completed by 738 individuals (74%). For the post-test in May 1998, 1016 individuals out of 8111, aged 18– 75 years, were selected at random. The distribution of questionnaires followed that of the pre-test, and 684 (68%) individuals completed this questionnaire. The advantage of the samples being representative of the population at the two time points motivated use of this procedure. Data on demographics and smoking habits are given in Table 1. According to Chi-square analyses, the two samples did not differ significantly with respect to any of the demographic variables, except for the number of children at home, which was higher in the post-test sample. This sample also included a higher proportion of smokers. Seventy-eight individuals participated in both surveys. The samples were drawn by Statistics ¨ rebro. Sweden in O The questions used in this study are given in Tables 2 (perceived pollution), 3 (risk perception), 4 (health risk perception; at posttest only for the path analysis), 5 (annoyance) and 6 (health symptoms; predominantly bronchitis- and asthma-like, and neurasthenic). For the factor ‘air pollution’ in the path analysis, the participants were categorized as being exposed to either a high or low level of pollution, depending on their residential area. High exposure was related to a location relatively close to the steel industry (Swedish post code areas 110, 111, 120, 124, 131 and 210 in
Table 1 Frequency (%) data on demographics and smoking habits for the pre- and post-test samples.
Gender Male Female Marital status Single Married or cohabiting Age (years) 18–29 30–44 45–64 >64 Highest level of education Elementary school High school University Occupation Employed Unemployed Student Retired Time lived in Oxelo¨sund (years) <1 1–5 >5 Time lived in present residence (years) 1–9 10–19 20–29 >29 Number of children at home 0 1–2 >2 Household member employed by SSAB or Oxelo¨sund Harbour Presently or formerly Never Smoking Presently/formerly Never **P<0.01; ***P<0.001; ns, not significant.
Pre-test
Post-test
c2
365 (49.5) 373 (50.5)
343 (50.1) 341 (49.9)
0.10 (ns)
157 (22.5) 541 (77.5)
157 (24.6) 480 (75.4)
0.82 (ns)
93 193 320 132
90 175 292 127
(13.2) (25.6) (42.7) (18.5)
0.28 (ns)
288 (39.9) 316 (43.8) 118 (16.3)
234 (35.9) 296 (45.4) 122 (18.7)
2.82 (ns)
436 58 32 173
400 52 37 162
(61.4) (8.0) (5.7) (24.9)
1.01 (ns)
6 (0.8) 46 (6.2) 686 (93.0)
8 (1.2) 35 (5.2) 636 (93.6)
1.12 (ns)
238 159 73 58
217 155 90 66
(41.1) (29.4) (17.0) (12.5)
3.55 (ns)
508 (68.8) 191 (25.9) 39 (5.3)
342 (50.0) 252 (36.8) 90 (13.2)
64.78***
423 (57.3) 315 (42.7)
407 (59.5) 277 (40.5)
0.74 (ns)
380 (51.5) 358 (48.5)
427 (62.4) 257 (37.6)
17.46***
(12.6) (26.2) (43.4) (17.8)
(62.4) (8.3) (4.6) (24.7)
(45.1) (30.1) (13.8) (11.0)
T. Stenlund et al. / Public Health 123 (2009) 339–345 Table 2 Questions and response frequencies (%) pertaining to perceived pollution for the pre- and post-test. Pre-test Rank the degree of environmental exposure/problems in Oxelo¨sund from 1 (most) to 9 (least) Noise 1–3 4–6 7–9 Dust and fallout 1–3 4–6 7–9 Road traffic 1–3 4–6 7–9 Air pollution other than dust and fallout 1–3 4–6 7–9 Effects on sea coast 1–3 4–6 7–9 Odours 1–3 4–6 7–9 Wood burning 1–3 4–6 7–9 Refuse handling 1–3 4–6 7–9 Other exposures/problems 1–3 4–6 7–9 If you have problems with dust or contamination, where does this occur?a, b Car Laundry House façade Windows Indoors (through ventilator or windows) Garden furniture Boat
Post-test
110 (19.6) 275 (48.9) 177 (31.5)
0.86 (ns)
550 (83.7) 59 (9.0) 48 (7.3)
472 (77.4) 90 (14.8) 48 (7.8)
10.44**
78 (13.3) 299 (51.0) 209 (35.7)
108 (19.4) 251 (45.2) 197 (35.4)
8.33**
375 (61.7) 164 (27.0) 69 (11.3)
366 (62.5) 159 (27.1) 61 (10.4)
0.30 (ns)
358 (59.1) 173 (28.5) 75 (12.4)
384 (65.2) 131 (22.2) 74 (12.6)
6.52 (ns)
216 (34.8) 279 (44.9) 126 (20.3)
183 (31.6) 273 (47.2) 123 (21.2)
1.51 (ns)
65 (11.2) 150 (26.0) 363 (62.8)
99 (18.1) 149 (27.3) 298 (54.6)
12.23***
107 (18.4) 189 (32.5) 286 (49.1)
92 (16.7) 167 (30.3) 293 (53.0)
1.76 (ns)
22 (28.6) 12 (15.6) 43 (55.8)
22 (17.1) 15 (11.6) 92 (71.3)
5.74 (ns)
308 122 276 495 286 442 230
(41.7) (16.5) (37.4) (67.1) (38.8) (59.9) (31.2)
327 132 328 459 264 426 257
(47.8) (19.3) (48.0) (67.1) (38.6) (62.3) (37.6)
Table 3 Questions and response frequencies (%) pertaining to risk perception for the preand post-test.
c2
106 (17.7) 297 (49.5) 197 (32.8)
5.53 (ns) 1.92 (ns) 15.77*** 0.01 (ns) 0.02 (ns) 0.73 (ns) 6.62 (ns)
**P<0.01; ***P<0.001; ns, not significant. a Representing perceived pollution in the path analysis. b Several alternatives could be chosen.
Oxelo¨sund), and low exposure was related to a location relatively distant from this industry (remaining post code areas in Oxelo¨sund). Chi-square analyses were computed to compare frequency distributions across response alternatives for the two test occasions. An alevel of 0.01 was chosen to account for the large number of statistical tests being conducted. Path analysis was used to test the model (Fig. 1), which is a form of structural equation modelling for testing and estimating causal relationships using a combination of statistical data and qualitative causal assumptions. This analysis encourages confirmatory rather than exploratory modelling. It starts with a hypothesis, represents it as a model, operationalizes the constructs, and tests the model. To test the model with the post-test data, the factor ‘air pollution’ comprised the variable ‘air pollution exposure’ that was categorized as high (residential location relatively close to the steel industry) or low (residential location relatively distant from the steel industry). Composite measures were used for perceived
341
Contamination of dust and soot is a major environmental and health problem in Oxelo¨sund Agree Doubtful Disagree Mass media’s description of Oxelo¨sund is too negative Agree Doubtful Disagree Oxelo¨sund is an appropriate area to live and work Agree totally Doubtful Disagree Financial considerations are generally given higher priority than environmental considerations in Oxelo¨sund Agree Doubtful Disagree Certain environmental disturbances must be accepted in order to retain and create new jobs in Oxelo¨sund Agree Doubtful Disagree Contamination of cars and boats is a major environmental problem in Oxelo¨sund Agree Doubtful Disagree Everyone has to waive one’s own convenience for a better urban environment Agree Doubtful Disagree The problem with air pollution in Oxelo¨sund is considerably exaggerated Agree Doubtful Disagree Environmental risks are exaggerated by the environmental organizations Agree Doubtful Disagree
Pre-test
Post-test
c2
637 (86.9) 57 (7.8) 39 (5.3)
487 (72.3) 109 (16.2) 78 (11.5)
46.13***
489 (67.3) 149 (20.5) 89 (12.2)
450 (67.2) 145 (21.6) 75 (11.2)
0.60 (ns)
529 (72.3) 157 (21.4) 46 (6.3)
504 (75.4) 121 (18.1) 43 (6.4)
2.58 (ns)
477 (65.5) 183 (25.1) 68 (9.4)
382 (56.8) 203 (30.2) 87 (12.9)
11.92**
397 (54.2) 165 (22.5) 170 (23.3)
360 (53.3) 170 (25.2) 145 (21.5)
1.47 (ns)
504 (68.7) 125 (17.0) 105 (14.3)
414 (61.3) 169 (25.0) 92 (13.7)
13.71**
539 (73.9) 123 (16.9) 67 (9.2)
451 (67.4) 138 (20.6) 80 (12.0)
7.47 (ns)
280 (38.1) 199 (27.1) 256 (34.8)
352 (52.5) 166 (24.7) 153 (22.8)
33.86***
200 (27.7) 220 (30.5) 302 (41.8)
205 (31.0) 249 (37.6) 208 (31.4)
16.91***
**P<0.01; ***P<0.001; ns, not significant.
pollution (seven items specified in Table 2; Cronbach‘s a ¼ 0.75), health risk perception (10 items specified in Table 4; a ¼ 0.73), annoyance (five items specified in Table 5; a ¼ 0.83) and health symptoms (five items specified in Table 6; a ¼ 0.67) that constituted the number of positive responses for these factors except for annoyance. This latter factor was represented by the mean scale value across items (0 ¼ not at all, 1 ¼ a little, 2 ¼ to some extent, 3 ¼ to a large extent). As the proposed model (Fig. 1) is saturated, i.e. includes all relationships between the variables, the model will fit the data perfectly. The aim was to find a more parsimonious model that would still fit the data. Therefore, the sample was randomly divided into two independent groups of equal size (n ¼ 342). Using one of these groups, the model was first tested and modified and then validated in the second group. Root mean squared error of approximation (RMSEA) and Benler’s comparative fit (CFI) were used as goodness-of-fit indices. An RMSEA value of 0.05 is indicative of a good fit, and a value of 0.08 indicates a reasonable fit.27 Divergence from an RMSEA value of 0.05 can be tested, and a non-significant P-value of close fit
342
T. Stenlund et al. / Public Health 123 (2009) 339–345
Table 4 Questions and response frequencies (%) representing health risk perception in path analysis for the post-test. Have public health conditions in Oxelo¨sund degenerated due to the external environment in the past 10 years? Yes No Do not know Have respiratory problems increased in the Oxelo¨sund population due to air pollution in the past 10 years? Yes No Do not know It is hazardous to health to live in Oxelo¨sund? Yes No Do not know Are the public health conditions in Oxelo¨sund due to the contamination of soot/dust much worse than in other communities of the same size? Yes No Do not know Do your relatives and friends have the opinion that it is hazardous to health to live in Oxelo¨sund? Yes No Do not know Is it the general opinion in Oxelo¨sund that it is hazardous to health to live in this community? Yes No Do not know Is it the general opinion outside Oxelo¨sund that this community is dirty and hazardous to health to live in? Yes No Do not know Are you worried that contamination with soot and dust will give you: Bronchitis? Yes No Do not know Asthma? Yes No Do not know Lung cancer? Yes No Do not know
Table 5 Questions and response frequencies (%) pertaining to annoyance for the pre- and post-test. Pre-test
158 (23.4) 330 (48.9) 187 (27.7)
191 (28.3) 291 (43.0) 194 (28.7) 115 (17.0) 409 (60.6) 151 (22.4)
247 (36.6) 298 (44.2) 129 (19.2)
192 (28.5) 346 (51.3) 136 (20.2)
314 (46.7) 212 (31.5) 147 (21.8)
434 (64.4) 110 (16.3) 130 (19.3)
193 (29.9) 328 (50.9) 124 (19.2) 260 (39.6) 285 (43.4) 111 (17.0) 229 (34.8) 294 (44.7) 135 (20.5)
(PCLOSE) indicates that the RMSEA value is not significantly different from 0.05. A CFI value of 0.95 has been suggested to represent a fairly good fit.28 Results Intervention study Perceived pollution The degree of exposure/problem of dust and fallout was ranked as significantly less in the post-test compared with the pre-test, whereas road traffic and wood burning were ranked as becoming more of a problem. No significant difference was found for noise, air pollution other than dust and fallout, effects on sea coast, odours, refuse handling or other exposures/problems between test occasions. When there was a problem with dust or contamination, house façades were more commonly reported as the affected site in the post-test than in the pre-test. No significant differences of this type over time were found for cars, laundry, windows, indoors, garden furniture or boats (Table 2).
To what extent have you been annoyed by the following exposures in the past 3 months?a Dust/soot from road traffic Not at all A little To some extent To a large extent Dust/soot from industry/workshops/ central boiling plants Not at all A little To some extent To a large extent Odours from industry/workshops Not at all A little To some extent To a large extent Exhausts from buses/trucks Not at all A little To some extent To a large extent Exhausts from cars Not at all A little To some extent To a large extent During the past months, have you experienced any of the following disturbances where you live? Difficulties opening windows due to outside noise Often (every week) Occasionally Never Difficulties falling asleep due to outside noise Often (every week) Occasionally Never Difficulties airing due to poor-quality outside air/odour Often (every week) Occasionally Never Difficulties airing due to outside contamination of dust/soot Often (every week) Occasionally Never Is air pollution so annoying that you have seriously considered moving? Is noise so annoying that you have seriously considered moving?
Post-test
c2
405 206 68 38
(56.5) 389 (59.2) (28.7) 187 (28.5) (9.5) 46 (7.0) (5.3) 35 (5.3)
203 177 123 217
(28.2) (24.6) (17.1) (30.1)
233 224 116 152
(32.1) 304 (48.2) 85.51*** (30.9) 195 (30.9) (16.0) 100 (15.8) (21.0) 32 (5.1)
472 172 46 32
(65.4) 444 (67.3) (23.8) 147 (22.3) (6.4) 46 (7.0) (4.4) 23 (3.5)
1.35 (ns)
397 232 66 32
(54.6) 405 (61.4) (31.9) 191 (28.9) (9.1) 40 (6.1) (4.4) 24 (3.6)
8.02 (ns)
24 (3.3) 22 (3.3) 120 (16.5) 110 (16.5) 584 (80.2) 535 (80.2)
0.01 (ns)
16 (2.2) 9 (1.3) 92 (12.7) 58 (8.7) 619 (85.1) 601 (90.0)
7.62 (ns)
285 163 108 103
3.06 (ns)
(43.2) 53.60*** (24.8) (16.4) (15.6)
45 (6.2) 24 (3.6) 11.09** 227 (31.2) 175 (26.2) 455 (62.6) 470 (70.2)
84 200 444 79
(11.5) 45 (6.8) 10.87** (27.5) 176 (26.5) (61.0) 444 (66.7) (10.7) 40 (5.8) 10.62**
21 (2.8)
26 (3.8)
0.79 (ns)
**P<0.01; ***P<0.001; ns, not significant. a Representing annoyance in the path analysis.
Risk perception In the post-test, the participants agreed to a lesser extent, compared with the pre-test, that contamination of dust and soot is a major environmental and health problem, that financial considerations are usually given higher priority than environmental considerations, and that contamination of cars and boats is a major environmental problem. Furthermore, they agreed to a greater extent in the post-test that the problem with air pollution is considerably exaggerated, and that environmental risks are exaggerated by the environmental organizations. No differences were found between test occasions regarding the mass media’s description, the appropriateness of the community as an area to
T. Stenlund et al. / Public Health 123 (2009) 339–345
343
Table 6 Questions and response frequencies (%) pertaining to health symptoms for the pre- and post-test. For comparison, corresponding percentages are given for neurasthenic and sensory symptoms as reference values for people living in multi-family (n ¼ 10,536) and single-family residences (n ¼ 3729).
Have you had any of the following symptoms in the past 12 months? Chest wheezing Tightness in chest Feeling of dyspnoea when waking up Waken up by coughing attacks Asthma attacks Have you had any of the following symptoms in the past 3 months?b Fatigue Feeling heavy headed Headache Nausea or dizziness Attentional difficulties Eye itching, burning or irritation Nose irritation, congestion or running Hoarseness or dry throat Coughing Face skin dryness or redness
c2
Pre-test
Post-test
130 83 49 173 24
(17.6) (11.2) (6.6) (23.4) (3.3)
129 80 47 188 23
(18.9) (11.7) (6.9) (27.5) (3.4)
0.31 (ns) 0.11 (ns) 0.04 (ns) 2.92 (ns) 0.01 (ns)
538 367 404 169 182 246 303 218 279 111
(72.9) (49.7) (54.7) (22.9) (24.7) (33.3) (41.1) (29.5) (37.8) (15.0)
502 360 488 158 175 245 305 196 276 124
(73.4) (52.6) (71.3) (23.1) (25.6) (35.8) (45.0) (28.7) (40.4) (18.1)
0.05 (ns) 1.14 (ns) 40.72*** 0.02 (ns) 0.13 (ns) 0.91 (ns) 1.94 (ns) 0.13 (ns) 0.95 (ns) 2.47 (ns)
Multi family residencec
Single-family residencec
65.9 45.2 49.8 18.3 21.5 28.4 41.2 31.5 32.9 19.2
62.7 37.4 47.8 12.7 15.3 19.4 34.2 21.6 27.1 11.8
a, b
**P<0.01; ***P<0.001; ns, not significant. a Representing symptoms in the path analysis. b Several alternatives could be chosen. c From Reference 31.
live and work, the acceptance of environmental disturbances to create new jobs, or waiving one’s own convenience for a better urban environment (Table 3). Annoyance Compared with the pre-test, dust or soot from industry, workshops or central boiler plants, and odours from industry or workshops were reported to be less annoying in the post-test, whereas no difference was found regarding dust or soot from road traffic or exhausts from buses, trucks or cars between occasions. Moreover, in the post-test, the participants experienced more frequent difficulties with airing because of poor-quality outside air or odour, as well as outside contamination of dust or soot. The test occasions did not differ with respect to difficulties in opening windows and falling asleep due to outside noise. Finally, the results found that fewer people were seriously considering moving due to air pollution, but not due to noise, in the post-test compared with the pretest (Table 5). Health symptoms In comparing the pre- and post-tests, no significant differences in prevalence were found for any types of pulmonary, neurasthenic or sensory symptoms, with the exception of headache which was reported as more prevalent in the post-test (Table 6). Testing the model Correlations between the variables in the model for each group are given in Table 7. As can be seen, all correlations with two exceptions were positive and significant. Using Amos 5.029 and the maximum likelihood method, the model was estimated for the first group. The analysis showed that with three exceptions, the path coefficients were significant (P < 0.05) and had the expected sign. The exceptions were the paths from air pollution to health risk perception and symptoms, and from perceived pollution to symptoms. The non-significant relationships were excluded and the modified model was re-estimated. In the modified model, all path coefficients were significant (P < 0.05). For the modified model, goodness-of-fit indices indicated a good fit (RMSEA ¼ 0.01, PCLOSE ¼ 0.68, CFI ¼ 0.99). Since the modified model was data driven to some extent, it should be validated on an independent
sample. The modified model was estimated for the second group and again all path coefficients were significant (P < 0.05) and the fit was good (RMSEA < 0.001, PCLOSE ¼ 0.93, CFI ¼ 1.00). The modified model and the validated model are shown in Fig. 2. Discussion Population-based questionnaire surveys were conducted in a Swedish community before and after an intervention that aimed to decrease air pollution (predominantly dust and soot, and also odorous substances). This consisted of closing a sinter plant and shipping pellets to the community’s harbour. As hypothesized, after the intervention, the environment was perceived as being less polluted by dust and soot, the residents were more positive in their risk perception (beliefs, attitudes and judgements), and they reported less annoyance due to dust, soot and odorous substances. In these respects, the intervention can be considered successful. Whereas less annoyance was reported after the intervention, the results did not support the hypothesis of an interventionassociated decrease in health symptoms, with respect to either pulmonary, neurasthenic or sensory symptoms. It should be noted that health symptoms typically become a problem at far higher exposure levels than annoyance.30 However, one can only speculate whether the air pollution, even prior to the intervention, was not prominent enough to evoke considerably more symptoms than is normal in Swedish communities. Reference data of this type for persons living in multi-family (n ¼ 10,536) and single-family residences (n ¼ 3729) are presented in Table 6 for neurasthenic and sensory symptoms.31 Although the symptom prevalences for Oxelo¨sund, both before and after the intervention, are generally Table 7 Correlation coefficients between air pollution (AP), perceived pollution (PP), health risk perception (HRP), annoyance (A) and symptoms (S) for the first/second path analysis.
PP HRP A S
AP
PP
HRP
A
0.29***/0.23*** 0.08 (ns)/0.12* 0.35***/0.45*** 0.01 (ns)/0.12*
0.40***/0.45*** 0.44***/0.39*** 0.14**/0.15**
0.42***/0.43*** 0.25***/0.34***
0.26***/0.26***
*P<0.05; **P<0.01; ***P<0.001; ns, not significant.
344
T. Stenlund et al. / Public Health 123 (2009) 339–345
Figure 2. Final path-analytic model of influences of air pollution, perceived pollution, health risk perception and annoyance on symptoms. Proportions of explained variance (r2) and standardized path coefficients are given for the first/second analysis.
somewhat higher than the reference prevalences, the difference is not large. Thus, this comparison may provide some support for the given speculation. Another possible explanation is that the larger percentage of smokers in the post-test sample contributed to pulmonary symptoms, which would overshadow effects of the intervention on these symptoms. There was also only a weak tendency of perceived pollution of odorous substances becoming weaker/less problematic after the intervention, which opposed the hypothesis. A contributing factor may be that pollution of odorous substances was not perceived as particularly strong/problematic even before the intervention. As would be expected from an intervention that aimed to specifically decrease exposure of dust, soot and odorous substances, neither perceived pollution nor annoyance due to environmental issues other than these were found to change after the intervention. This suggests that the differences regarding dust, soot and odorous substances between the two studied samples are not likely to be explained by differences in response behaviour. The two samples were also similar with respect to all studied demographic variables except for the number of children at home, which was larger in the post-test sample. This may have contributed to increased perceived pollution,22 and possibly to a more negative risk perception in this sample, which would imply that the effect of the intervention on perception of pollution and risk may have been even larger than suggested by the present data. It is worth noting that the neurasthenic symptoms (fatigue, feeling heavy headed, headache, nausea or dizziness, and attention difficulties) were rather high in prevalence, both before and after the intervention, compared with reference values, but this was less so for sensory symptoms (Table 6). It is not yet known to what extent the high prevalence of neurasthenic symptoms, both before and after the intervention, can be related to environmental exposure. The categorization of being exposed to either a high or low level of pollution was based on the assumption that residential areas closer to the steel industry were exposed to higher levels of pollution than areas further away from the industry. Although this may seem reasonable, it calls for certain caution since no chemical exposure data are available. It is possible that the absence of a relationship between air pollution and symptoms can be related to this assumption not being fulfilled. A second objective of this study was to empirically test and further develop a model of air pollution and its psychological and medical consequences. The path analyses suggest that air pollution influences perceived pollution which, in turn, influences health risk perception and annoyance. There is also a direct effect of air pollution on annoyance. Health risk perception has an effect on annoyance and symptoms, whereas annoyance influences symptoms. These findings suggest that perceived pollution and health risk perception do indeed play important roles in understanding
and predicting environmentally induced annoyance and health symptoms. In particular, this is the case for health risk perception which explains a rather large proportion of the variance in outcome variables. However, rather small proportions of the variance in perceived pollution and symptoms are explained by the factors in the model, which calls for certain caution. It is likely that various health issues not included in this study would explain considerable variance in symptoms. It is important to point out that the direction of effects in the proposed model in Fig. 2 calls for caution. Thus, the path analysis does not test the cause–effect direction between factors per se. Instead, the directions in the model are causal assumptions based on theory and/or prior empirics. However, the causal assumptions in this model can be questioned. For example, in contrast to the model, it has been suggested that an association between perceived pollution and health risk perception may be due to health risk perception affecting perceived pollution.23 Hence, it cannot be excluded that perceived pollution and health risk perception mutually affect each other, and that symptoms can affect health risk perception and annoyance. It is also quite possible that if people are annoyed by air pollution, they may take measures to reduce this pollution, although in this particular setting, it may be difficult for individuals to reduce pollution. It is important to point out that the cause–effect directions in the proposed model are the most likely directions. In conclusion, perceived pollution, annoyance attributed to dust, soot and odorous substances, and risk perception were reported to be less pronounced after the intervention compared with prior to the intervention. Path analyses further suggest that perceived pollution and health risk perception play important roles in understanding and predicting environmentally induced annoyance and health symptoms. Acknowledgement The authors wish to thank Svenskt Stål Aktiebolag (SSAB), ¨ rebro, and the EnvironOxelo¨sunds hamn, Statistics Sweden in O ment and Protection Board in Oxelo¨sund for their valuable support, and Inger Fagerlund for excellent assistance. Ethical approval Swedish Data Inspection Board. Funding Grants from the Swedish Environmental Protection Agency and ¨ rebro County Council, Sweden. the Environmental County Fund, O Competing interests None declared. References 1. Theophanides M, Anastassopoulou J, Vasilakos C, Maggos T, Theophanides T. Mortality and pollution in several greek cities. J Environ Sci Health Part A 2007;42:741–6. 2. Pervin T, Gerdtham UG, Lyttkens CH. Societal costs of air pollution-related health hazards: a review of methods and results. Cost Eff Resour Alloc 2008;6:19. 3. Frost K, Frank E, Maibach E. Relative risk in the news media: a quantification of misrepresentation. Am J Public Health 1997;87:842–5. ¨ sterberg K, O ¨ stergren PO. Prevalance of 4. Carlsson F, Karlsson B, Orbaek P, O annoyance attributed to electrical equipment and smells in a Swedish population, and relationship with subjective health and daily function. Public Health 2005;119:568–77. 5. Pope CA. What do epidemiologic findings tell us about health effects of environmental aerosols? J Aerosol Med 2000;13:335–54. 6. Leikauf GD. Hazardous air pollutants and asthma. Environ Health Perspect 2002;110(Suppl. 4):505–26.
T. Stenlund et al. / Public Health 123 (2009) 339–345 7. Berglund B, Berglund U, Lindvall T. Measurement and control of annoyance. In: Koelga HS, editor. Environmental annoyance: characterization, measurement and control. Amsterdam: Elsevier Science; 1987. p. 29–43. 8. Neher JO, Koenig JQ. Health effects of outdoor air pollution. Am Fam Physician 1994;49:1397–404. 9. Chen TM, Gokhale J, Shofer S, Kuschner WG. Outdoor air pollution: nitrogen dioxide, sulphur dioxide and carbon monoxide health effects. Am J Med Sci 2007;333:249–56. 10. Paz C. Some consequences of ozone exposure on health. Arch Med Res 1997;28:163–70. 11. Nattero G, Enrico A. Outdoor pollution and headache. Headache 1996;36:243–5. 12. Strøm-Tejsen P, Weschler CJ, Wargocki P, Mysko´w D, Zarzycka J. The influence of ozone on self-evaluation of symptoms in a simulated aircraft cabin. J Expo Sci Environ Epidemiol 2008;18:272–81. 13. Kampa M, Castanas E. Human health effects of air pollution. Environ Pollut 2008;151:362–7. 14. Sunyer J, Jarvis D, Gotschi T, Garcia-Esteban R, Jacquemin B, Aguilera I, et al. Chronic bronchitis and urban air pollution in an international study. Occup Environ Med 2006;63:836–43. 15. Maynard RL. Asthma and urban air pollution. Clin Exp Allergy 2001;31:518–20. 16. Oglesby L, Kunzli N, Monn C, Schindler C, Ackermann-Liebrich U, Leuenberger P. Validity of annoyance scores for estimation of long term air pollution exposure in epidemiologic studies: the Swiss Study on Air Pollution and Lung Diseases in Adults (SAPALDIA). Am J Epidemiol 2000;152:75–83. 17. Elliott SJ, Cole DC, Krueger P, Voorberg N, Wakefield S. The power of perception: health risk attributed to air pollution in an urban industrial neighbourhood. Risk Analysis 1999;19:621–34. 18. Pidgeon NF, Hood C, Jones D, Turner BA, Gibson R. Risk perception. In: Risk analysis, perception and management. London: Royal Society; 1992. p. 89–134. 19. Oxelo¨sund Municipality. Slutrapport PDS 1989-06-01 [Final report from ¨ sunds kommun; 1989. a problem detection study]. Oxelo 20. Bickerstaff K. Risk perception research: socio-cultural perspectives on the public experience of air pollution. Environ Int 2004;30:827–40.
345
21. Petrie KJ, Sivertsen B, Hysing M, Broadbent E, Moss-Morris R, Eriksen HR, et al. Thoroughly modern worries: the relationship of worries about modernity to reported symptoms, health and medical care utilization. J Psychosom Res 2001;51:395–401. 22. Hunter PR, Bickerstaff K, Davies MA. Potential sources of bias in the use of individuals recall of the frequency of exposure to air pollution for use in exposure assessment in epidemiological studies: a cross-sectional survey. Environ Health 2004;31:3. 23. Hunter PR, Davies MA, Hill K, Whittaker M, Sufi F. The prevalence of selfreported symptoms of respiratory disease and community belief about the severity of pollution from various sources. Int J Environ Health Res 2003;13:227–38. 24. Lipscomb JA, Satin KP, Neutra RR. Reported symptom prevalence rates from comparison populations in community-based environmental studies. Arch Environ Health 1992;47:263–9. 25. Williams ID, McCrae IS. Road traffic nuisance in residential and commercial areas. Sci Total Environ 1995;169:75–82. 26. Butani L, O’Connell EJ. Functional respiratory disorders. Ann Allergy Asthma Immunol 1997;79:91–9. 27. Brown MW, Cudeck R. Alternative ways of assessing model fit. In: Bollen KA, Long JS, editors. Testing structural equation models. New Park, CA: Sage; 1993. p. 136–62. 28. Hu L-T, Bentler PM. Cutoff criteria for fit indexes on covariance structure analysis: conventional criteria versus new alternatives. Struct Equat Model 1999;6:1–55. 29. Arbuckle JL, Wothke W. Amos 4.0 user’s guide. Chicago: Small Waters Corporation; 1999. 30. Baird JC, Berglund B, Berglund U, Lindvall T. Symptom patterns as an early warning signal of community health. Environ Int 1990;16:3–9. 31. Andersson K, Norle´n U. Indoor climate and health effects. The indoor climate in the Swedish housing stock. In: Flatheim G, Berg K, Edvardsen K, editors. Indoor air quality in practice. Moisture and cold climate solutions; 1995. p. 23–8. Oslo: Norwegian Society of Chartered Engineers.