- Page 4-113, lines 13 through 19 focuses on CVM, but really is about stated preferencemethods, including CVM applications. Some <strong>of</strong> <strong>the</strong> past economic studies are more like SPchoice studies than conventional CVM applications. Consequently, it may be appropriate tomerge <strong>the</strong> paragraphs starting on lines 13 and 20.- Page 4-113, line 19 could use a citation, ei<strong>the</strong>r NAPAP or Chestnut and Dennis, or Mitchelland Carson.- Does <strong>the</strong> Hanley and Spash reference discuss visibility applications in specific and in detail?- Page 4-113, line 31. “Davis” should be “Dennis”.Health Risk Assessment (Staff paper Chapter 4 and separate paper).- I support conducting <strong>the</strong> assessment in more than 2 locations, as discussed at <strong>the</strong> meeting.- Staff paper 4-13, lines 10-26 discusses assumptions about changes in ambient conditions tomeet standards, relying predominately on <strong>the</strong> rollback method. Using <strong>the</strong> rollback method isreasonable, but EPA should give careful attention to <strong>the</strong> proposed sensitivity analysis <strong>of</strong>alternative adjustments (lines 24-26). With increasing costs <strong>of</strong> compliance, episodic ando<strong>the</strong>r control strategies that reduce <strong>the</strong> highest concentrations may receive increasedattention. Fur<strong>the</strong>r, given that <strong>the</strong> population exposed is not uni<strong>for</strong>m across concentrationlevels, and many concentration-response functions are non-linear, differences in <strong>the</strong>assumptions to reduce concentrations to achieve standards can have a significant impact on<strong>the</strong> risk assessment.- Deck et al, 2001 is cited several times, starting in <strong>the</strong> first paragraph, but is not available. Itmay be useful to provide this paper <strong>for</strong> this review.<strong>Criteria</strong> <strong>Document</strong> Chapter 9This chapter is well done as a series <strong>of</strong> separate summaries, but it needs more integrationand needs to be reduced in length – not everything needs to be summarized. It appropriatelyfocuses on <strong>the</strong> larger questions <strong>of</strong> increasing consistency in <strong>the</strong> results <strong>of</strong> available health effectsliterature and extensions to this literature. In terms <strong>of</strong> <strong>the</strong> important question <strong>of</strong> retaining orrevising <strong>the</strong> existing PM 2.5 standard levels (15 ug/m3 annual average and 65 ug/m3 24 hours),little is presented in this chapter on <strong>the</strong> strength <strong>of</strong> <strong>the</strong> evidence, shapes <strong>of</strong> <strong>the</strong> estimated C-Rfunctions around <strong>the</strong>se levels, or effect thresholds (although this is touched on in Section 6.4.6).Jonathan Samet, MDChapter 5 - General Comments:In general, this is a cohesive and thorough chapter that carefully sets out concepts <strong>of</strong> exposureassessment, measurement approaches, and findings. The literature review appears complete andfindings are well represented in tables and in <strong>the</strong> text. The chapter has a potentially key role insetting a framework <strong>for</strong> interpreting <strong>the</strong> epidemiological data presented in Chapter 6. Thechapter does address <strong>the</strong> implications <strong>of</strong> <strong>the</strong> exposure literature <strong>for</strong> interpreting <strong>the</strong>epidemiological evidence. Un<strong>for</strong>tunately, <strong>the</strong>re is little linkage between <strong>the</strong> two chapters in thisregard; Chapter 6 almost reads as though Chapter 5 had not preceded it. There is a need <strong>for</strong>better integration, a burden which clearly lies with <strong>the</strong> authors <strong>of</strong> Chapter 6.This chapter also discusses issues related to confounding and measurement error that overlapwith Chapter 6. With regard to issues <strong>of</strong> confounding, it will be important to have a uni<strong>for</strong>mview throughout <strong>the</strong> CD. My comments <strong>for</strong> Chapter 6 should be considered in this regard.A - 73
Specific Comments:Page 5-19, lines 15-18: Generalizability (external validity) is not dealt with well here. A“purposeful study” may give generalizable in<strong>for</strong>mation; <strong>the</strong> extent <strong>of</strong> generalizability is a matter<strong>of</strong> judgment, based on study participant characteristics. The term statistical inference is usedinappropriately here.Page 5-41, lines 5-16: This paragraph is far too sweeping in its condemnation <strong>of</strong> <strong>the</strong> exposureassessment literature. What are <strong>the</strong> “important questions” that are so poorly answered.Chapter 6 - General Comments:This lengthy chapter provides an exhaustive, descriptive summary <strong>of</strong> <strong>the</strong> most recentepidemiological findings on particulate matter and morbidity and mortality. The literaturereview is comprehensive and <strong>the</strong> tables <strong>of</strong>fer useful summaries <strong>of</strong> an extensive literature. Thereare, however, weaknesses that should be addressed; <strong>the</strong>se weaknesses may reflect <strong>the</strong> multiauthorednature <strong>of</strong> <strong>the</strong> chapter which has resulted in an uneven approach in style, syn<strong>the</strong>sis, andinterpretation. Key aspects <strong>of</strong> <strong>the</strong> chapter needing to be addressed include:1. The chapter is not adequately connected to <strong>the</strong> remainder <strong>of</strong> <strong>the</strong> CD. There is a lack <strong>of</strong>integration with Chapter 5, which should provide a foundation <strong>for</strong> exposure considerationsrelated to interpretation <strong>of</strong> <strong>the</strong> epidemiological literature. This foundation is not used, and farless strong and competent text is provided.2. The chapter fails to sharply set out key concepts—confounding, causal associations, andcausal pathways, in particular. Effect modification is also not handled well and <strong>the</strong> text relatedto <strong>the</strong>se key aspects <strong>of</strong> interpretation is <strong>of</strong>ten murky. In <strong>the</strong> chapter’s introduction, it would beuseful to provide diagrams indicating <strong>the</strong> relationships that hold under confounding, direct causalpathways, and indirect causal pathways. I have attached one possible set <strong>of</strong> diagrams.Additionally considerations as to confounding, reflect biological understanding as to <strong>the</strong>independent action <strong>of</strong> <strong>the</strong> confounder and not just patterns <strong>of</strong> association in data. Changes inestimates are not a particularly useful gauge as to <strong>the</strong> presence <strong>of</strong> confounding in <strong>the</strong> presence <strong>of</strong>measurement error, possible effect modification, and correlations among <strong>the</strong> independentvariables. The text in places <strong>of</strong>fers some clear thinking on <strong>the</strong>se difficult topics, but much <strong>of</strong> it isnot clear.3. In interpreting data, <strong>the</strong>re is excessive reliance on p values and attaining statisticalsignificance <strong>for</strong> effect estimates. This needs to be corrected; <strong>the</strong> p value should not be held as adecision-making tool <strong>for</strong> data interpretation, as in <strong>the</strong> present chapter (see Epidemiology <strong>for</strong> arecent set <strong>of</strong> perspectives on this issue. Emphasis should be placed instead on precision <strong>of</strong> effectestimates (i.e., width <strong>of</strong> confidence intervals..4. In fact, <strong>the</strong> chapter never clearly sets out how data will be interpreted. Summary judgmentsare <strong>of</strong>fered but without reference to any common framework Judgments are <strong>of</strong>ten couchedsubjectively and <strong>the</strong>re is a general failure to place <strong>the</strong> epidemiological evidence within <strong>the</strong>broader context <strong>of</strong> toxicological and dosimetric understanding. The framework <strong>for</strong> interpretationis badly needed. Throughout <strong>the</strong> text, <strong>the</strong>re is variably use <strong>of</strong> significance level, precision <strong>of</strong>effect estimates, and magnitude <strong>of</strong> effect, as <strong>the</strong> effects <strong>of</strong> PM are weighed against those <strong>of</strong> o<strong>the</strong>rpollutants. The same problem is evident when <strong>the</strong> chapter interprets <strong>the</strong> literature on particlecharacteristics and components.Specific CommentsPage 6-2, lines 11-15: This proposed hierarchy <strong>of</strong> “inferential strength” is nei<strong>the</strong>r correct noruseful.Page 6-3, lines 19-21: This type <strong>of</strong> sweeping statement should be avoided. On close reading,<strong>the</strong> sentence <strong>of</strong>fers only a garbled pejorative comment.A - 74
- Page 5:
known, the potential causes deserve
- Page 10:
SCIENCE ADVISORY BOARD STAFFMr. A.
- Page 13 and 14:
Page 2-77, line 19-22: Should menti
- Page 15 and 16:
Page 8-1, lines 26-28: Combustion a
- Page 17 and 18:
mode vs. the other. In fact, such k
- Page 19 and 20:
p. 7-49, l. 20 In an effort to make
- Page 21 and 22:
ambient PM effects. The paragraph d
- Page 23 and 24:
dominant, one criticism is that “
- Page 25 and 26:
6. Susceptible sub-populationsIt is
- Page 27 and 28:
is OK. But in most settings it stil
- Page 29 and 30:
2. There are repetitions of the sam
- Page 31 and 32:
15. Page 3-57 and 3-58, line 29-31
- Page 33 and 34: P 7- 27, L 15: What does “compara
- Page 35 and 36: there is pertains almost solely to
- Page 37 and 38: studies, and is presented as observ
- Page 39 and 40: P 8-47, L 23-27: These two sentence
- Page 41 and 42: also be summarized. Second, the cha
- Page 43 and 44: P 9-76, L 30: It should be “these
- Page 45 and 46: 4. Page 4-7,lines 14-18. Similar th
- Page 47 and 48: control when it may be possible to
- Page 49 and 50: 1990. Reference Lioy, P.J. “The A
- Page 51 and 52: document.P. 5-82, Lines 15-30 Need
- Page 53 and 54: 7-12 8 insert "that are either very
- Page 55 and 56: 8-62 10,11 The preceding discussion
- Page 57 and 58: 9-27 17 insert "source and/or" afte
- Page 59 and 60: 2. The paper by Künzli et al. on t
- Page 61 and 62: 6-243 12 This section (6.4.4.) shou
- Page 63 and 64: Chapter 5. Human Exposure to PM and
- Page 65 and 66: the chapter. Many of the poor quali
- Page 67 and 68: Page 5-17, equation 5-10; the coeff
- Page 69 and 70: illustrated using a figure from Kel
- Page 71 and 72: tied back to the base-line health s
- Page 73 and 74: Page 7-4, Structure of the Respirat
- Page 75 and 76: Günter Oberdörster, PhDChapter 7
- Page 77 and 78: efficiencies as well as the ratio o
- Page 79 and 80: The title of this section is also s
- Page 81 and 82: old and young rats and mice used on
- Page 83: passive use values as opposed to us
- Page 87 and 88: Chapter 9 - General CommentsThis ch
- Page 89 and 90: George Taylor, PhDAir Quality Crite
- Page 91 and 92: atmospheric stressors associated wi
- Page 93 and 94: “At the surface, a variable fract
- Page 95 and 96: that point from this review! “Vis
- Page 97 and 98: 6. P 2-86, section 2.2.5.1 - A shor