Table 4.

Path Interventions (Type B).

Authors	Intervention & Study		N, Response Rate & Method	Exposure Levels		Change in Levels and Distribution of Change across Participants	Outcome Measure(s) before and after Outcomes	Did Outcome Change with Change in Exposure? Yes/No (Significance Tested?)	Before/after Outcome Change Compared to That Estimated from an ERF	Comments	Confounders Adjusted for in Analyses
	Nature	Design		Before	After
Amundsen, Klaeboe & Aasvang (2011) [27]	Norway Façade insulation	Two survey rounds: B&A Target; Control & Supplement groups. B & A surveys approx 6 mos. either side of intervention	Target: B: 168 Response rates 57% A: 161 (65%) Control: B: 469 (57%) A: 254 (65%) Supplement: 112 Mail survey	LAeq,24 h 61–78 outside. Calculated Mean 71 (Av inside level before change is 43)	Façade insulation reduces inside levels by 7 dB	–7 dB for indoor noise levels for all Ps in target group	Standard ISO annoyance scale (5 point verbal). %HA calculated using top two points of scale B: 42%HA A: 16%HA Control: B: 24%HA A: 29%HA	Yes Intervention resulted in substantial and significant (p < 0.001) reductions in individual annoyance scores	Authors chose not to compare their results with Miedema & Oudshoorn ERF [13]. Fitted a model of individual annoyance responses to outdoor levels for all Ps (target, control and supplementary: n = 738) with receiving the intervention as a dummy variable. Estimate of effect size −0.820 (p < 0.000) and 95% CI −1.170 to −0.470. Authors claim size of annoyance reductions with intervention is in line with ERF modelled from individual indoor levels. However this appears to be contradicted by the large reduction in the Target Group’s %HA (42% before intervention to 16% after). Summarised as ‘unclear’	Authors note no explanation why %HA sig. lower in control than target before intervention; and second round higher than first in control	Gender, age, education level, marital status, access to a bedroom on the quiet side of the building, and sensitivity
Amundsen, Klaeboe & Aasvang (2013) [28]	See Amundsen, Klaeboe & Aasvang (2011) [27] above. Same study details but this was a repeat survey 2 year after first post-intervention study. Mailed to all Ps who had completed first post-intervention study. Number of participants now 104 (Response rate 58%) in target; 139 in Control; 63 in supplement						2nd after-study: A: 16%HA	Result the same two years after initial After survey (p < 0.01) Additionally, repeated ANOVA was conducted on panel who had answered all three survey rounds (N = 212). Change in annoyance as a result of intervention significant in first (p < 0.0005) but not second (p = 0.33) after survey	In this repeated ANOVA, multivariate partial eta square = 0.44
Bendtsen, Michelsen & Christensen (2011) [29]	Denmark Enlargement of motorway lanes but with dwelling insulation, barriers, & quiet pavement	B/A study 1 year before constr & 1 year after B/A gap 6 year	Q sent to 1200 dwel. In 6 areas out to800 m from motorway Response rates B:71% A: 65% 38% B&A Mail survey	Lden 45–65 Calculated. Unclear as to whether calculated levels included traffic sources other than motorway	Lden 45–60 Calculated. Not reported is whether some Ps may have experience increased after-levels	Reductions in extent of exposure 60–65 & 55–60 bands but increase in lower two bands. Reported only at population level. No indication of the change experienced by individual Ps	ISO scale (5 point verbal) % top three annoyance categories dropped, other two categories increased Top two categories (%HA—but authors did not use this term) dropped from 37% to 16%	Yes but no data presented of change in exposure of those reporting change in annoyance No statistical tests	n.a.	Classed as path intervention, even though includes quiet pavement as intervention Multiple sources of road traffic exposure—not just motorway
Gidlöf-Gunnarsson, Öhrström & Kihlman (2010) [30]	Sweden Full-scale filling-in building gaps; barriers & housing improvement	B/A study 5 year apart	B: 160 Response rate: 56% A: 153 (47%) Mail survey	LAeq,24 h at façade 48–71 Calculated		–5 to −10 on exposed facades; −4 to −10 courtyards	ISO scale (5 point verbal) %Annoyed cut-off includes top 3 points.(Note: NOT %HA) For Ps highly exposed and with large change: B: 84% Annoyed A: 28% Annoyed For P’s with less change: B: 45–55% Annoyed A: 21–22% Annoyed	Yes Large and consistent reductions in %Annoyed associated with reduction in noise exposure (but no statistical tests)	Authors refer to Öhrström [7] who cites ERF of Miedema & Vos [12] For Ps in most exposed part of study, B/A 84/28%Annoyed outcomes both higher than estimated by this ERF, but also show much larger decrease in response than estimated by ERF. Response to change in same direction as estimated by ERF, but steeper, indicating excess response. (But no statistical tests)	This was a reconstruction project that included many other environmental changes—not just change in noise exposure (Ps reported 36% increase in overall satisfaction with area)
Kastka, Buchta, Ritterstaedt, Paulsen & Mau (1995) [31]	Germany Noise barriers at 12 sites + 2 control sites	B/A study 1–2 years B, 8–10 years A barriers were built	283 B Response rate 59% 212 A (72%) 97 Ps both B&A	Leq,D B 50–70 Measured	Leq,D A:51–66 Measured	–13 dB close to barriers to 0 dB at 200 m Av. Change −4.1 dB	(1) 5 point verbally labelled disturbance scale (2) %HA calc. as top two responses on scale in (1) (3) factor K1: sensory-perceptional and emotional experience of traffic noise (0–10) (4) factor K2: noise interferences	Yes All response variables show significant reductions, e.g., %HA B: 64%; A:35% (chi2 = 39.69 p < 0.005) Control sites response variables same B&A	Authors calculated an ERF using the steady-state before-responses. For this, mean disturbance scores (and, separately, other outcome variables including %HA) were regressed against before Leq,D Mean (After) disturbance scores were compared to those estimated by this ERF. At 11 of the 12 sites, estimated mean disturbance score was greater than observed. Difference was statistically significant (matched pair t-test, df = 11, p < 0.05). Response to change in same direction as estimated by ERF, but steeper, indicating excess response following barrier construction	Authors reported extensive additional analyses They suggest no simple causal relation between noise level reduction and annoyance reduction
Nilsson & Berglund (2006) [32]	Sweden Noise barrier	B/A study + control 9 mos. B; 15 mos. A Repeated measures on 59%, 46% only	Before 304 Response rate 77% (241 control Response rate 66%) (After Response rates: 72%, 69%) Self-administered	Lden 70 to <45 Calculated	Lden 62.5 to <45 Calculated	–7.5 dB; with reducing change out to 100 m from barrier. Distribtn of change was: –7.5 dB 52 Ps −5 dB 47 Ps −2.5 dB 31 Ps	Visual analogue scale 7-point annoyance scale. Transformed to 0–100 scale. Reports %HA as above cut-off 72	Yes Reductions in %HA were significant (p < 0.05, sign-test) for three groups of Ps within 100 m of roadway Control: no diff in B&A %HA	ERF cited was Miedema & Oudshoorn (2001) [13] Reports both B&A %HA agree with prediction by ERF (no statistical test) Response to change same direction and magnitude as estimated by ERF	Outdoor annoyance did not conform to ERF
Vincent & Champelovier (1993) [33]	France Noise barriers and low noise road surface	B/A study at 2(?) sites.	75 Response rate not reported	Leq,12 h 65.1 Location of measurement site relative to Ps not reported	Leq,12 h 56.3 Location not reported	Change in levels was variable with distance from road: −10 to −3 dB between 10 and 100 m.	% highly annoyed (scale and definition of HA not reported). B: 22%HA A: 8%HA	Yes (but no statistical test)	No comparison of change to any ERF	Author notes that response to ‘Often disturbs sleep’ dropped from 13% to 6%