ABCDEFGHIJKLMNOPQRSTUVWXYZAA
1
Literature Search for RCTs on Clean Cookstoves and Health Outcomes, Spring 2018
2
Indicates a study or component outcome not included in GiveWell's interim intervention report on Clean Cookstoves
3
NOTE: some studies are associated with multiple papers that include results for different outcomes, and in these cases we have consolidated related papers in one column
4
CAPS (Cooking and Pneumonia Study)Hanna et al. 2016Katz et al. 2016; Tielsch et al. 2016GRAPHS (Ghana Randomized Air Pollution and Health Study)Alexander et al. 2017RESPIRE (Randomised Exposure Study of Pollution Indoors and Respiratory Effects)Hartinger et al. 2016Kirby et al. 2019Romieu et al. 2009Bensch & Peters 2012 (working paper)Aung et al. 2018
5
Location:MalawiIndiaNepalGhanaNigeriaGuatemalaPeru RwandaRural MexicoSenegalIndia
6
Sample size:10,750 enrolled, 10,543 reporting (Main health study)
6168 enrolled, 4754 analyzed (School health attendance results)		2,575 households in 44 villages2,553 pregnancies in 3376 households enrolled (Birth outcomes results)
5254 children from 3376 households enrolled (Health outcomes results)		1,414 enrolled, 1,313 analyzed (Birth outcomes results)
324 (BP results)
271 (Inflammation results)		534 households503 households 1,582 households552 women253 households222
7
8
Results on women's health:
9
Women's blood pressuren/aNo statistically significant differencen/aNon-randomized evidence: Transient BP increases immediately following acute smoke exposureStatistically significant reduction in diastolic BP (-2.8 mm Hg) but not systolic BP. 6.4% of control subjects were hypertensive (SBP ≥140 and/or DBP ≥90 mm Hg) versus 1.9% of ethanol users (P = 0.051).Non-randomized evidence: Modest BP decreases that reach significance for diastolic BP: 3.7 mm Hg lower SBP [95% confidence interval (CI), −8.1 to 0.6] and 3.0 mm Hg lower DBP (95% CI, −5.7 to −0.4) after controlling for confounders and a measured decrease in exposure in the intervention group.n/an/an/an/aBP goes the wrong direction for non-exclusive users of the new stoves: SBP 2.6 (−0.4, 5.7) mmHg) and DBP (1.2 (−0.9, 3.3) mmHg). In TOT effects on exclusive stove users, there are small insignificant BP reductions: SBP (−2.0 (−4.5, 0.5) mmHg) and DBP (−1.1 (−2.9, 0.6) mmHg).
10
Women's lung function (spirometry test)n/aNo statistically significant differencen/an/an/a−0.02, 95% CI (−0.09, 0.04). Goes the wrong direction and not statistically significant.n/an/aIn TOT results, lung function declined by half as much (31 ml) compared with the open fire use (62 ml) over 1 year of follow-up (P = 0.012) for women 20 years of age and older. No statistically significant difference in ITT results, and endline worse than baseline for both groups. n/an/a
11
Other health symptomsn/aNo statistically significant difference in health index (-0.008), inc. probability of cough, any illness in last 30 days (self-reported)n/an/aNo overall difference for 5 tested biomarkers of inflammation in pregnant women. 1 of 5 became statistically significant in a subgroup analysis. Statistically significant reduction in wheeze, 1 of 6 assessed symptoms. RR was slightly lower in intervention group on remaining symptoms, but not close to significance. n/an/aStatistically significant reduction in 11 self-reported health symptoms on a TOT basis adjusted for controls. Significant 10 percentage point reduction in eye irritation for at least 1 woman responsible for cooking (1 of 4 self-reported health outcomes). 8 pp reduction in at least one woman showing reduction in symptoms of respiratory disease borderline significant at the 10% level. Statistically significant ITT reductions in self-reported eye irritation
12
13
Results on children's health
14
Children's PneumoniaIntervention versus control incidence rate ratio (IRR) of pneumonia cases in under-5s was 1·01 (95% CI 0·91–1·13; p=0·80)n/a13% decline in mother-reported acute lower respiratory infections incidence up to age 36 mos, but the evidence was weak (RR 0·87, 95% CI 0·67–1·13). No decline in severe acute lower respiratory infections. Significant reduction in bacterial but not viral pneumonia in treatment group that received gas stoves (unclear whether there's an overall reduction in cases)n/aRR of 0·84, 95% CI 0·63–1·13; p=0·257 for all physician-diagnosed pneumonia cases from birth up to 18 months old. Significant decrease in physician-diagnosed severe pneumonia cases [RR 0·67 (0·45–0·98) p=0·042]. Similar results for fieldworker-assessed pneumonia cases. No statistically significant difference in acute lower respiratory infections, ages 6-35 months. The study found a statistically significant decline in caregiver-reported acute respiratory infections in children under 5 over the last 7 days (prevalence ratio [PR] 0.75, 95% CI 0.60–0.93, p = 0.009). However, households also received water treatment interventions, and the study found no significant impact on 48-hour personal exposure to log-transformed fine particulate matter (PM2.5) concentrations among cooks (β = −0.089, p = 0.486) or children (β = −0.228, p = 0.127).n/an/an/a
15
Children's birthweightn/a
No statistically significant difference, estimate negative (self-reported)
No significant difference in 'adverse birth outcomes' (The paper is somewhat vague, but I believe this refers to mean birthweight, LBW, preterm, and small-for-gestational age.)No statistically significant difference in either armn/an/an/an/an/an/an/a
16
Other health outcomesCooking-related serious adverse events (burns) were seen in 19 children; nine in the intervention and ten (one death) in the control group (IRR 0·91 [95% CI 0·37–2·23]; p=0·83). School attendance: school absenteeism was unchanged for children 5-18 in treatment communities [IRR 0.92 (0.71–1.18), p = 0.51].
No statistically significant difference in health index (0.006) (self-reported)
Modest, signficant improvements in persistent cough (0·91, 0·85–0·97), wheeze (0·87, 0·78–0·97) and burn injury (0·68, 0·48–0·95). No improvements in fever or ear discharge. No statistically significant difference low birthweight or preterm in either armn/an/aNo statistically significant difference in child growth (ages 6-35 months). Declines in reported diarrhea that were not statistically significant (relative rate of 0.78 [95% confidence interval (CI): 0.58–1.05]).The intervention reduced the prevalence of reported child diarrhea by 29% (prevalence ratio [PR] 0.71, 95% confidence interval [CI] 0.59–0.87, p = 0.001)n/an/an/a
17
Links to source(s)Main health outcome results are from Mortimer et al. 2017: https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(16)32507-7/fulltext

School attendance results: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0193376		Hanna et al. 2016, Tables 6A, p. 104 and 6B, p. 105: https://www.povertyactionlab.org/sites/default/files/publications/52%20Up%20in%20Smoke%20AEJ2016.pdfResults on children's health outcomes:
https://www.researchgate.net/publication/300080651/download

Results on birth outcomes: https://www.researchgate.net/publication/306351476_Impact_of_an_improved_biomass_stove_on_birth_outcomes_in_rural_Nepal_A_cluster-randomized_step-wedge_trial		Birth outcomes from Wylie 2017: https://www.ajog.org/article/S0002-9378(16)31902-0/pdf

Pneumonia results from Lee et al. 2017: https://www.atsjournals.org/doi/abs/10.1164/ajrccm-conference.2017.195.1_MeetingAbstracts.A5976

Women's blood pressure results from Quinn et al. 2017: https://ehjournal.biomedcentral.com/articles/10.1186/s12940-017-0282-9		Blood pressure results are from Alexander et al. 2017: https://www.atsjournals.org/doi/abs/10.1164/rccm.201606-1177OC

Inflammation results are from Olopade et al. 2017: https://www.sciencedirect.com/science/article/pii/S0160412016307358		Pneumonia results are from Smith et al. 2011, pp. 1717, 1719, and Table 2, p. 1722: http://cleancookstoves.org/resources_files/eff-ect-of-reduction-in.pdf

Women's respiratory function and symptoms results are from Smith-Sivertsen et al. 2009, Figure 4 & Table 2: https://academic.oup.com/aje/article/170/2/211/110882

Women's blood pressure results are from McCracken et al. 2007: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1913602/		https://academic.oup.com/ije/article/45/6/2089/2452363
https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1002812
Romieu et al. 2009, Abstract and Table 3: https://www.atsjournals.org/doi/abs/10.1164/rccm.200810-1556OC?url_ver=Z39.88-2003&rfr_id=ori%3Arid%3Acrossref.org&rfr_dat=cr_pub%3DpubmedBensch and Peters 2012, P. 6, Table 3 p. 20, & Table 9 p. 27: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2030746https://www.sciencedirect.com/science/article/pii/S0013935117314949
18
Notes on trial design features:Large community RCT of a cleaner biomass stove compared to open fire cooking. It has the largest sample size for which data were collected of all trials we are aware of. Field randomized controlled trial without intensive clinical oversight ("typical use"). Measured reduction in exposure to air pollution was only statistically significant for women in year 1 of 4.RCT of improved biomass stoves that led to a measured improvement in measurements of air pollution. However, the authors note that they nevertheless remained much higher than recommended levels. This RCT included two treatment arms, one that received reduced emissions biomass stove and a second that received gas stoves. It is notable that this is one of a limited number of trials of gas cookstoves on health outcomes. Notable for the treatment being gas stoves rather than improved biomass and for having the strongest randomized trial results on blood pressure.Efficacy-focused randomized controlled trial of chimney cookstoves. Featured intensive weekly follow-up and quality assessment of stove function and use, so its results probably do not reflect "typical use" of the intervention without strong monitoring. Combines improved cookstoves with sanitation interventions, so the effect of the cookstove intervention is not isolated. This study is a large community RCT that treated roughly 101,000 households overall, though it surveyed a much smaller random sample at endline. It combined improved cookstoves with water treatment interventions, so any effects of cookstoves are not isolated. All health outcomes were caregiver-reported.

Because the intervention was multifaceted and the RCT did not find evidence of reduced exposure to air pollution, it's difficult to attribute the potential health impacts to clean cookstoves in particular. 		RCT of chimney stoves, but adherence was low (50%), and the authors used a treatment-on-the-treated analysis (results limited to stove users in the treatment group) and added control variables. Stove adopters are subject to selection bias, so the reported results aren't random. We're also unsure if this method of analysis was pre-registered, so we have concerns about p-hacking. Small trial that doesn't seem to be published. It reports relatively high stove adoption (~75% of total stove usage from cleaner sources). All health outcomes are self-reported. The main reported health impact of reduced eye irritation doesn't seem like a major benefit. It would be interesting to know if anything about this trial or the stoves used led to higher-than-normal adoption. Small trial. Notes lower-than-expected efficiency of the intervention stove at reducing air pollution.
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100