Chapter  91:  Health Effects of Omega-3 Fatty Acids on Asthma

Metabolism and Biological Effects of Essential Fatty Acids

Dietary fat is an important source of energy for biological activities in human beings. It encompasses saturated fatty acids, which are usually solid at room temperature, and unsaturated fatty acids, which are liquid at room temperature. Unsaturated fatty acids can be further divided into monounsaturated and polyunsaturated fatty acids. Polyunsaturated fatty acids (or PUFAs) can be classified, on the basis of their chemical structure, into two groups: omega-3 (n-3) fatty acids and omega-6 (n-6) fatty acids. The omega-3 or n-3 notation means that the first double bond in this family of PUFAs is 3 carbons from the methyl end of the molecule. The same principle applies to the omega-6 or n-6 notation. Despite their differences in structure, all fats contain the same amount of energy (i.e., 9 kcal/g or 37 kJ/g).

Of all fats found in food, two—alpha-linolenic acid (chemical abbreviation: ALA; 18:3 n-3) and linoleic acid (LA; 18:2 n-6)—cannot be synthesized in the human body, yet these are necessary for proper physiological functioning. These two fats are thus called “essential fatty acids.” The essential fatty acids can be converted in the liver to long-chain polyunsaturated fatty acids (LC PUFAs), which have a higher number of carbon atoms and double bonds. These LC PUFAs retain the omega type (n-3 or n-6) of the parent essential fatty acids.

ALA and LA comprise the bulk of the total PUFAs consumed in a typical North American diet. Typically, LA comprises 89% of the total PUFAs consumed, while ALA comprises 9%. Smaller amounts of other PUFAs make up the remainder.¹ Both ALA and LA are present in a variety of foods. For example, LA is present in high concentrations in many commonly used oils, including safflower, sunflower, soy, and corn oil. ALA, which is consumed in smaller quantities, is present in leafy green vegetables and in some commonly used oils, including canola and soybean oil. Some novelty oils, such as flaxseed oil, contain relatively high concentrations of ALA, but these oils are not commonly found in the food supply.

The Institute of Medicine (IOM) suggests that, for adults 19 and older, an adequate intake (AI) of ALA is 1.1–1.6 grams/day, and 11–17 grams/day for LA.² Recommendations regarding AI differ by age and gender groups, and for special conditions such as pregnancy and lactation.

Figure 1. Classical omega-3 and omega-6 fatty acid (more...)

   Figure 1. Classical omega-3 and omega-6 fatty acid synthesis pathways and the role of omega-3 fatty acids in regulating health/disease markers

Figure 1

Metabolic Pathways of Omega-3 and Omega-6 Fatty Acids

Figure 1

The specific biological functions of fatty acids depend on the number and position of double bonds and the length of the acyl chain. Both EPA and AA are 20-carbon fatty acids and are precursors for the formation of prostaglandins (PGs), thromboxane (Tx), and leukotrienes (LTs)—hormone-like agents that are members of a larger family of substances called eicosanoids. Eicosanoids are localized tissue hormones that seem to be one of the fundamental regulatory classes of molecule in most higher forms of life. They do not travel in the blood, but are created in the cells to regulate a large number of processes, including the movement of calcium and other substances into and out of cells, dilation and contraction of muscles, inhibition and promotion of clotting, regulation of secretions including digestive juices and hormones, and, the control of fertility, cell division, and growth.⁴

Figure 1

In addition to affecting eicosanoid production as described above, EPA also affects lipoprotein metabolism and decreases the production of other compounds—including cytokines, interleukin 1β (IL-1β and tumor necrosis factor a (TNF-a)—which have pro-inflammatoryeffects. These compounds exert pro-inflammatory cellular actions that include stimulating the production of collagenase and increasing the expression of adhesion molecules necessary for leukocyte extravasation.⁶ The mechanism responsible for the suppression of cytokine production by omega-3 LC PUFAs remains unknown, although suppression of eicosanoid production by omega-3 fatty acids may be involved. EPA can also be converted into the longer chain omega-3 form of docosapentaenoic acid (DPA, 22:5 n-3), and then further elongated and oxygenated into DHA. EPA and DHA are frequently referred to as VLN-3FA—very long chain n-3 fatty acids. DHA, which is thought to be important for brain development and functioning, is present in significant amounts in a variety of food products, including fish, fish liver oils, fish eggs, and organ meats. Similarly, AA can convert into an omega-6 form of DPA.

Studies have reported that omega-3 fatty acids decrease triglycerides (Tg) and very low density lipoprotein (VLDL) in hypertriglyceridemic subjects, concomitant with an increase in high density lipoprotein (HDL). However, they appear to increase or have no effect on low density lipoprotein (LDL). Omega-3 fatty acids apparently lower Tg by inhibiting VLDL and apolipoprotein B-100 synthesis, and decreasing post-prandial lipemia.⁷ Omega-3 fatty acids, in conjunction with transcription factors (small proteins that bind to the regulatory domains of genes), target the genes governing cellular Tg production and those activating oxidation of excess fatty acids in the liver. Inhibition of fatty acid synthesis and increased fatty acid catabolism reduce the amount of substrate available for Tg production.⁸

As noted earlier, omega-6 fatty acids are consumed in larger quantities (> 10 times) than omega-3 fatty acids. Maintaining a sufficient intake of omega-3 fatty acids is particularly important since many of the body's physiologic properties depend upon their availability and metabolism.

U.S. Population Intake of Omega-3 Fatty Acids

The major source of omega-3 fatty acids is dietary intake of fish, fish oil, vegetable oils (principally canola and soybean), some nuts such as walnuts, and, dietary supplements. Two population-based surveys, the third National Health and Nutrition Examination (NHANES III) 1988-94, and the Continuing Survey of Food Intakes by Individuals 1994-98 (CSFII), are the main sources of dietary intake data for the U.S. population. NHANES III collected information on the U.S. population aged =2 months. Mexican Americans and non-Hispanic African-Americans, children =5 years old, and adults = 60 years old were over-sampled to produce more precise estimates for these population groups. There were no imputations for missing 24-hour dietary recall data. A total of 29,105 participants had complete and reliable dietary recall.

The CSFII 1994-96, popularly known as the “What We Eat in America” survey, addressed the requirements of the National Nutrition Monitoring and Related Research Act of 1990 (Public Law 101-445) for continuous monitoring of the dietary status of the American population. The CSFII 1994-96 utilized an improved data-collection method for 24-hour recall known as the multiple-pass approach. Given the large variation in intake from day-to-day, multiple 24-hour recalls are considered to be best suited for most nutrition monitoring and will produce stable estimates of mean nutrient intake from groups of individuals.⁹ In 1998, the Supplemental Children's Survey, a survey of food and nutrient intake by children under the age of 10 years, was conducted as a supplement to the CSFII 1994-96. The CSFII 1994-96, 1998 surveyed 20,607 people of all ages with over-sampling of low-income population (<130% of the poverty threshold). Dietary intake data from individuals of all ages were collected over 2 nonconsecutive days via two 1-day dietary recalls.

Table 1. Estimates of the mean±standard error (more...)

Table 1. Estimates of the mean±standard error of the mean (SEM) intake of linoleic acid (LA), alpha-linolenic acid (ALA), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA) in the US population, based on analyses of a single 24-hour dietary recall of NHANES III data

	Grams/day		% Kcal/day
	Mean±SEM	Median (range)1	Mean±SEM	Median (range)1
LA (18:2 n-6)	14.1±0.2	9.9 (0 – 168)	5.79±0.05	5.30 (0 – 39.4)
ALA (18:3 n-3)	1.33±0.02	0.90 (0 – 17)	0.55±0.004	0.48 (0 – 4.98)
EPA (20:5 n-3)	0.04±0.003	0.00 (0 – 4.1)	0.02±0.001	0.00 (0 – 0.61)
DHA (22:6 n-3)	0.07±0.004	0.00 (0 – 7.8)	0.03±0.002	0.00 (0 – 2.86)

¹The distributions are not adjusted for the over-sampling of Mexican-Americans, non-Hispanic African-Americans, children =5 years old, and adults = 60 years old in the NHANES III dataset.

Table 2. Mean, range, median, and standard error of (more...)

Table 2. Mean, range, median, and standard error of the mean (SEM) of usual daily intakes of linoleic acid (LA), total omega-3 fatty acids (n-3 FA), alpha-linolenic acid (ALA), eicosapentaenoic acid (EPA), docosapentaenoic acid (DPA) and docosahexaenoic acid (DHA) in the US population, based on CSFII data (1994-1996, 1998)

	Grams/day
	Mean±SEM	Median±SEM
LA (18:2 n-6)	13.0±0.1	12.0±0.1
Total n-3 FA	1.40±0.01	1.30±0.01
ALA (18:3 n-3)	1.30±0.01	1.21±0.01
EPA (20:5 n-3)	0.028	0.004
DPA (22:5 n-3)	0.013	0.005
DHA (22:6 n-3)	0.057±0.018	0.046±0.013

Dietary Sources of Omega-3 Fatty Acids

Table 3. The omega-3 fatty acid content, in grams per (more...)

Table 3. The omega-3 fatty acid content, in grams per 100 g food serving, of a representative sample of commonly consumed fish, shellfish, fish oils, nuts and seeds, and plant oils that contain at least 5 g omega-3 fatty acids per 100 g

Food item	EPA	DHA	ALA
Fish (Rawa)
Anchovy, European	0.6	0.9	-
Bass, Freshwater, Mixed Sp.	0.2	0.4	0.1
Bass, Striped	0.2	0.6	trace
Bluefish	0.2	0.5	-
Carp	0.2	0.1	0.3
Catfish, Channel	trace	0.2	0.1
Cod, Atlantic	trace	0.1	trace
Cod, Pacific	trace	0.1	trace
Eel, Mixed Sp.	trace	trace	0.4
Flounder & Sole Sp.	trace	0.1	trace
Grouper, Mixed Sp.	trace	0.2	trace
Haddock	trace	0.1	trace
Halibut, Atlantic and Pacific	trace	0.3	trace
Halibut, Greenland	0.5	0.4	trace
Herring, Atlantic	0.7	0.9	0.1
Herring, Pacific	1.0	0.7	trace
Mackerel, Atlantic	0.9	1.4	0.2
Mackerel, Pacific and Jack	0.6	0.9	trace
Mullet, Striped	0.2	0.1	trace
Ocean Perch, Atlantic	trace	0.2	trace
Pike, Northern	trace	trace	trace
Pike, Walleye	trace	0.2	trace
Pollock, Atlantic	trace	0.4	-
Pompano, Florida	0.2	0.4	-
Roughy, Orange	trace	-	trace
Salmon, Atlantic, Farmed	0.6	1.3	trace
Salmon, Atlantic, Wild	0.3	1.1	0.3
Salmon, Chinook	1.0	0.9	trace
Salmon, Chinook, Smokedb	0.2	0.3	-
Salmon, Chum	0.2	0.4	trace
Salmon, Coho, Farmed	0.4	0.8	trace
Salmon, Coho, Wild	0.4	0.7	0.2
Salmon, Pink	0.4	0.6	trace
Salmon, Pink, Cannedc	0.9	0.8	trace
Salmon, Sockeye	0.6	0.7	trace
Sardine, Atlantic, Canned in Oild	0.5	0.5	0.5
Seabass, Mixed Sp.	0.2	0.4	-
Seatrout, Mixed Sp.	0.2	0.2	trace
Shad, American	1.1	1.3	0.2
Shark, Mixed Sp.	0.3	0.5	trace
Snapper, Mixed Sp.	trace	0.3	trace
Swordfish	0.1	0.5	0.2
Trout, Mixed Sp.	0.2	0.5	0.2
Trout, Rainbow, Farmed	0.3	0.7	trace
Trout, Rainbow, Wild	0.2	0.4	0.1
Tuna, Fresh, Bluefin	0.3	0.9	-
Tuna, Fresh, Skipjack	trace	0.2	-
Tuna, Fresh, Yellowfin	trace	0.2	trace
Tuna, Light, Canned in Oile	trace	0.1	trace
Tuna, Light, Canned in Watere	trace	0.2	trace
Tuna, White, Canned in Oile	trace	0.2	0.2
Tuna, White, Canned in Watere	0.2	0.6	trace
Whitefish, Mixed Sp.	0.3	0.9	0.2
Whitefish, Mixed Sp., Smoked	trace	0.2	-
Wolffish, Atlantic	0.4	0.3	trace
Shellfish (Raw)
Abalone, Mixed Sp.	trace	-	-
Clam, Mixed Sp.	trace	trace	trace
Crab, Blue	0.2	0.2	-
Crayfish, Mixed Sp., Farmed	trace	0.1	trace
Lobster, Northern	-	-	-
Mussel, Blue	0.2	0.3	trace
Oyster, Eastern, Farmed	0.2	0.2	trace
Oyster, Eastern, Wild	0.3	0.3	trace
Oyster, Pacific	0.4	0.3	trace
Scallop, Mixed Sp.	trace	0.1	-
Shrimp, Mixed Sp.	0.3	0.2	trace
Squid, Mixed Sp.	0.1	0.3	trace
Fish Oils
Cod Liver Oil	6.9	11.0	0.9
Herring Oil	6.3	4.2	0.8
Menhaden Oil	13.2	8.6	1.5
Salmon Oil	13.0	18.2	1.1
Sardine Oil	10.1	10.7	1.3
Nuts and Seeds
Butternuts, Dried	-	-	8.7
Flaxseed			18.1
Walnuts, English	-	-	9.1
Plant Oils
Canola (Rapeseed)	-	-	9.3
Flaxseed Oil	-	-	53.3
Soybean Lecithin Oil	-	-	5.1
Soybean Oil	-	-	6.8
Walnut Oil	-	-	10.4
Wheatgerm Oil	-	-	6.9

Trace = <0.1; - = 0 or no data; Sp. = species;

a

Except as indicated;

b

Lox.;

c

Solids with bone and liquid;

d

Drained solids with bone;

e

Drained solids.

Asthma: A Chronic Inflammatory Disease

The National Heart, Lung, and Blood Institute (NHLBI) defines asthma as follows:

Asthmatic episodes are triggered by a variety of stimuli including allergens, environmental irritants, viral infections, exercise or other poorly defined factors.

Overview

The UO-EPC's evidence report on omega-3 fatty acids and asthma is based on a systematic review of the scientific-medical literature to identify, and synthesize the results from, studies addressing key questions. Together with content experts, UO-EPC staff identified specific issues integral to the review. A Technical Expert Panel (TEP) refined the research questions, as well as highlighted key variables requiring consideration in the evidence synthesis. Evidence tables presenting the key study characteristics and results were developed. Summary tables were derived from the evidence tables. The methodological quality of the included studies was appraised, and individual study results were summarized.

Key Questions Addressed in This Report

The purpose of this evidence report was to synthesize information from relevant studies to address the following seven questions:

• What is the evidence for the efficacy of omega-3 fatty acids to improve respiratory outcomes among individuals with asthma? (Question 1)

• What is the evidence that the possible value (efficacy/association) of omega-3 fatty acids in improving respiratory outcomes is dependent on the: specific type of fatty acid (DHA, EPA, DPA, ALA, fish, fish oil); specific source (fish, plant, food, dietary supplement [fish oil, plant oil]); its serving size or dose (fish or dietary supplement); amount/dose of omega-6 fatty acids given as a cointervention; ratio of omega-6/omega-3 fatty acids used; fatty acid content of blood lipid biomarkers; absolute fatty acid content of the baseline diet; relative fatty acid content of the baseline diet; tissue ratios of fatty acid (omega-6/omega-3) during the investigative period; intervention length; anti-oxidant use; and, the manufacturer and its product(s) (purity; presence of other potentially active agents)? (Question 2)

• What is the evidence that, in individuals with asthma, omega-3 fatty acids influence mediators of inflammation which are thought to be related to the pathogenesis of asthma? (Question 3)

• Are omega-3 fatty acids effective in the primary prevention of asthma? (Question 4)

• Among individuals with asthma, do omega-3 fatty acids alter the progression of asthma (i.e., secondary prevention)? (Question 5)

• What is the evidence for adverse events, side effects, or counter-indications associated with omega-3 fatty acid use to treat or prevent asthma (DHA, EPA, DPA, ALA, fish oil, fish)? (Question 6)

• What is the evidence that omega-3 fatty acids are associated with adverse events in specific subpopulations of asthmatic individual such as diabetics? (Question 7)

Four questions (1, 2, 3, and 5) concern treatment or secondary prevention, one centers on primary prevention (4), and two focus on adverse events, side effects, or counter-indications (6, 7).

Analytic Framework

Figure 2. Analytic Framework for omega-3 fatty acids (more...)

   Figure 2. Analytic Framework for omega-3 fatty acids in asthma. Populations of interest in rectangles. Exposure in oval. Outcomes in rounded rectangles. Effect modifiers in hexagons. Solid connecting arrows indicate associations and effects reviewed in this report

Figure 2

Not all associations within the analytic framework were investigated. Regarding those individuals with asthma, the key questions focused on the impact of the omega-3 fatty acid intervention/exposure on:

• respiratory outcomes (Question 1);

• mediators of inflammation thought to play a key role in the pathogenesis of asthma (Question 3);

• the progression of asthma (Question 5); and,

• the likelihood of adverse events, side effects, or counter-indications (Questions 6 and 7).

Another question focused on whether or not covariates (e.g., omega-3 fatty acid type or source; omega-6 fatty acid intake as a cointervention) could account for the observed effect on respiratory outcomes (Question 2). The level of fatty acids in the human body, for example the fatty acid content in phospholipids of cell membranes of polymorphonuclear or mononuclear leukocytes, was also investigated insofar as it could act as an effect modifier with respect to respiratory outcomes (Question 2). While it could be assumed that a positive influence on respiratory outcomes may result from the exposure's effect on mediators of inflammation (e.g., leukotrienes; Question 3), the direct association between mediators of inflammation and respiratory outcomes was not assessed.

For questions relating to treatment efficacy, the primary respiratory outcome was forced expiratory volume in one second (FEV₁), considered by many to be the gold standard measure of respiratory functioning. This decision was made in consultation with our TEP. Secondary respiratory outcomes assessed ultimately depended on what outcomes were measured in the included studies. For questions related to secondary prevention, a longterm perspective on respiratory functioning is required.

The question relating to primary prevention (Question 4) involves two populations: those with an elevated risk of developing asthma and those healthy individuals who may develop it. Of primary significance to this question is the prevalence or incidence of asthma as well as its severity. Investigation of both the primary and secondary prevention questions could conceivably include examining the links involving effect modifiers (e.g., risk factors) or mediators of inflammation.

Study Identification

Data Abstraction

Following a calibration exercise involving two studies, three reviewers independently abstracted the contents of each included study using an electronic Data Abstraction form developed especially for this review (Appendix C). The studies were divided evenly, with half assigned to each of two reviewers. Once a reviewer completed their work, they then checked all of the data abstracted by their counterpart. Data abstracted included the characteristics of the:

• report (e.g., publication status, language of publication, year of publication);

• study (e.g., sample size; research design; number of arms, cohorts, or phases; funding source);

• population (e.g., age; percent males; diagnosis description, including severity, duration, and concomitants; comorbid conditions);

• intervention/exposure (e.g., omega-3 fatty acid types, sources, doses, and intervention/exposure length), and comparator(s);

• cointerventions (e.g., asthma medications, omega-6 fatty acids); and,

• withdrawals and dropouts.

Data relating to outcomes (i.e., respiratory, mediators of inflammation, safety) and two covariates (fatty acid content of blood lipid biomarkers, tissue ratios of fatty acid during the investigative period) were abstracted by a third reviewer, all of whose work was then checked by one of the first two reviewers.

Summarizing the Evidence

Results of Literature Search

Regardless of its source, the progress of each bibliographic record through the stages of the systematic review is illustrated in the modified QUOROM flow chart (Appendix D). Ideally, a record included an abstract and key words, in addition to a citation. When a citation was discovered, for example through a manual search of a reference list, its complete bibliographic record was sought (e.g., Pubmed) and then entered into the first level of relevance screening.

Of 1,010 records entered into the initial screening for relevance, 851 were excluded. Reflecting the specific eligibility criteria, the reasons for exclusion were: a. not a primary study (e.g., a review; n = 246); b. does not involve human participants (n = 170); c. does not involve omega-3 fatty acids as an exposure/intervention (n = 250); and, d. the purpose of the exposure/intervention was not the treatment or prevention of asthma (n = 185). All but five of the remaining 159 reports were then retrieved and subjected to a more detailed relevance assessment.^42–46 One report was retrieved, but it was not translated in time to include it in the systematic review.⁴⁷ The second relevance screening then excluded 122 reports for the following reasons: a. not a primary study (e.g., a review; n = 70); b. does not involve human participants (n = 4); c. does not involve omega-3 fatty acids as an exposure/intervention (n = 14); and, d. the purpose of the exposure/intervention was not the treatment or prevention of asthma (n = 34). In total, 31 reports, describing 26 unique studies, were deemed relevant for the systematic review, with five studies each described by two reports.

Two reports referred to Huang et al.'s primary prevention (observational) study.^{48, 49} A published preliminary report⁵⁰ outlined the protocol for Mihrshahi et al.'s longterm primary prevention RCT whose 18-month results were recently published.⁵¹ The results of Hodge et al.'s treatment RCT⁵² were first disseminated in an abstract;⁵³ and, Kirsch et al.'s treatment RCT had its clinical outcome data reported in one publication⁵⁴ and its mediators of inflammation data in another.⁵⁵ Finally, two published reports of a treatment RCT appeared to overlap substantially. When the lead author of the Arm et al. reports was contacted he confirmed that the two reports described a single RCT, although one of them⁵⁶ provided response to allergen-challenge data from the participants described in the previous publication,⁵⁷ as well as from several other subjects. These data had been unavailable for inclusion in the earlier report, in which clinical data and some mediators of inflammation data are presented. To avoid entering duplicate study participant data, we followed the author's recommendation and focused on the first report⁵⁷ for everything except the response to allergen-challenge data, which we obtained from the second document.⁵⁶ To avoid confusion in the text, tables, or figures, only one report is used to refer to a given study and its data. It is the one reporting the most data pertaining to the study. Some information regarding the study design of an RCT exclusively described by an abstract⁵⁸ were taken from the Cochrane review,³⁵ which had obtained additional details from a source unavailable to the present review team.

Report and Study Design Characteristics of Included Studies

Of the included studies, two were abstracts^{53, 58} and the rest were articles published in scientific journals. Only one study was not described by at least one published report.⁵⁸ The one relevant study identified by manual search was published.⁵⁹ All but five reports (all published) were written in English. Two required translation from Russian^{60, 61} two from Japanese,^{47, 62} and one from Polish.⁶³ As reported earlier, one Japanese publication was retrieved for the purposes of assessing its relevance, yet it was not translated in time to include it in this report.⁴⁷ Given its abstract, which allowed it to pass the initial relevance screening, it appears to have been a non-RCT examining the effects of EPA on asthma symptoms, fatty acids in serum, and the generation of various leukotrienes in response to leukocytes.

RCTs and less rigorous types of design were found to address five of the seven research questions. The latter were either controlled (i.e., non-RCT) or uncontrolled (i.e., noncomparative case series; cohort study).

Summary Table 1: RCT evidence of omega-3 fatty acids (more...)

Summary Table 1: RCT evidence of omega-3 fatty acids to improve respiratory outcomes in asthma

Author, Year, Location	Omega-3 Fatty Acid Arm/Phase		Comparator Arm/Phase		# of S Unique Results Favoring Omega-3 Fatty Acids	Jadad Total Quality / Allocation Concealment (Internal Validity)	Applicability (External Validity)
	n	Type & Dose/Day	n	Type & Dose/Day
ADULTS
Arm, 1988, England56,57	NR	3.2 g EPA + 2.2 g DHA	NR	Olive oil (NR)	1/9	4 (Grade: A)/unclear	III
Emelyanov, 2002, Russia69	23	200 mg EPA+DHA + 400 mg olive oil	23	600 mg olive oil	4/7	5 (Grade: A)/ unclear	III
Kirsch, 1988, USA54,55	6	4.0 g EPA ethyl ester (trace DHA)	6	0.1 g EPA ethyl ester (trace DHA)	0/6	3 (Grade: B)/unclear	I
McDonald, 1991, Australia58*	15	2.7 g EPA + 1.8 g DHA	15	15 g olive oil	0/4	3 (Grade: B)/unclear	III
Okamoto, 2000a, Japan66	7	10–20 g perilla seed oil (ALA: NR)	7	10–20 g corn oil	4/4	2 (Grade: C)/unclear	III
Stenius-Aarniala, 1989, Finland68*	36	20 mL fish oil (EPA+DHA: NR)	36	(1) 20 mL olive oil	0/2	2 (Grade: C)/unclear	III
				(2) 20 mL evening primrose oil
Thien, 1993, England67	NR	3.2 g EPA + 2.2 g DHA	NR	Olive oil (NR)	0/6	4 (Grade: A)/unclear	III
CHILDREN
Hodge, 1998, Australia52	NR	0.72 g EPA 0.48 g DHA + ALA (NR) via canola diet	NR	Omega-6 fatty acids: 1.8 g safflower oil + 1.8 g palm oil + 0.4 g olive oil + sunflower diet (NR)	0/3	3 (Grade: B)/unclear	III
Nagakura, 2000, Japan64	15	17.0–26.8 mg/kg EPA; 7.3–11.5 mg/kg DHA (300 mg fish oil)	15	300 mg olive oil	2/2	4 (Grade: A)/unclear	III
NOT REPORTED
Dry, 1991, France65	NR	1.0 g EPA+DHA	NR	“Placebo” (NR)	1/1	2 (Grade: C)/unclear	X

n = number of enrolled participants; NR = not reported; S = significant;

*

Crossover trial

Summary Table 2: Evidence from other study designs (more...)

Summary Table 2: Evidence from other study designs of omega-3 fatty acids to improve respiratory outcomes in asthma

Author, Year, Location	Omega-3 Fatty Acid Cohort/Phase		Comparator Cohart/Phase		# of S Unique Results Favoring Omega-3 Fatty Acids	Total Quality (Internal Validity)	Applicability (External Validity)
	n	Type & Dose/Day	n	Type & Dose/Day
ADULTS
Ashida, 1997, Japan59†	5	15 g perilla seed oil (ALA: NR)	NA	NA	3/3	3 (Grade: B)	III
Broughton, 1997, USA73†	26	“Low” EPA + DHA intake: ~0.7 g (mean)	26	“High” EPA + DHA intake: ~3.3 g (mean)	5/16	3 (Grade: B)	I
Hashimoto, 1997, Japan62†	8	1.8 g EPA	NA	NA	5/7	3 (Grade: B)	III
Masuev, 1997b, Russia61††	5	6.0 g EPA+DHA	3	6.0 g olive oil	1/1	4 (Grade: A)	III
Okamoto, 2000b, Japan71†	26	10–20 g of perilla seed oil (ALA: NR)	NA	NA	5/5	4 (Grade: A)	III
Picado, 1988, Spain74†	10	3.0 g EPA+DHA + eucaloric diet	10	3.0 g lactose + eucaloric diet	3/4	5 (Grade: A)	III
Villani, 1998, Italy72†	7	3.0 g EPA+DHA (1:1 ratio)	NA	NA	3/7	4 (Grade: A)	III
CHILDREN
Gorelova, 1998, Russia70††	23	4.5 g fish oil (NR) + hypo-allergenic diet	10	Control (NR) + hypo-allergenic diet	1/1	2 (Grade: C)	III
Machura, 1996, Poland63††	37	3.0 g EPA+DHA (15 mL fish oil)	23	15 mL sunflower oil	2/6	3 (Grade: B)	III

n = number of enrolled participants; NR = not reported; NA = not applicable; S = significant;

†

Noncomparative case series;

††

non-RCT

Summary Table 3: An indirect assessment of the impact (more...)

Summary Table 3: An indirect assessment of the impact of effect modifiers on the value of omega-3 fatty acids to improve four respiratory outcomes in asthma, organized by research design

	Omega-3 Fatty Acid Arm/Phase		Comparator Arm/Phase
Author, Year, Location	n	Type & Dose/Day	n	Type & Dose/Day	Results Favoring Omega-3 Fatty Acids	Research Design	Applicability (External Validity)
ADULTS
Arm, 1988, England56,57	NR	3.2 g EPA + 2.2 g DHA	NR	Olive oil (NR)	NS: AM PEF	RCT	III
					NS: PM PEF
					NS: BDU
Dry, 1991, France65*	NR	1.0 g EPA+DHA	NR	“Placebo” (NR)	S: FEV1	RCT	X
Emelyanov, 2002, Russia69	23	200 mg EPA+DHA + 400 mg olive oil	23	600 mg olive oil	S: AM PEF	RCT	III
					S: BDU
					NS: FEV1
					NS: PM PEF
Kirsch, 1988, USA54,55	6	4.0 g EPA ethyl ester (trace DHA)	6	0.1 g EPA ethyl ester (trace DHA)	NS: FEV1	RCT	I
McDonald, 1991, Australia58	15	2.7 g EPA + 1.8 g DHA	15	15 g olive oil	NS: AM PEF	RCT, 2-phase crossover	III
					NS: PM PEF
					NS: BDU
Okamoto, 2000a, Japan66	7	10–20 g perilla seed oil (ALA: NR)	7	10–20 g corn oil	S: FEV1	RCT	III
					S: AM PEF
Stenius-Aarniala, 1989, Finland68	36	20 mL fish oil (EPA+DHA: NR)	36	(1) 20 mL olive oil	NS: AM PEF	RCT, 3-phase crossover	III
				(2) 20 mL evening primrose oil	NS: PM PEF
Thien, 1993, England67	NR	3.2 g EPA + 2.2 g DHA	NR	Olive oil (NR)	NS: AM PEF	RCT	III
					NS: PM PEF
					NS: BDU
Ashida, 1997, Japan59	5	15 g perilla seed oil (ALA: NR)	NA	NA	S: AM PEF	Non-comparative case series	III
					S: PM PEF
Hashimoto, 1997, Japan62	8	1.8 g EPA	NA	NA	S: AM PEF	Non-comparative case series	III
Picado, 1988, Spain74	10	3.0 g EPA+DHA + eucaloric diet	10	3.0 g lactose + eucaloric diet	S: AM PEF	Non-comparative case series (placebo, then fish oil)	III
					NS: BDU†
Villani, 1998, Italy72	7	3.0 g EPA+DHA (1:1 ratio)	NA	NA	NS: AM PEF	Non-comparative case series	III

n = number of enrolled/randomized participants; NR = not reported; NA = not applicable; S = significant; NS = nonsignificant; BDU = bronchodilator use;

*

Age not reported

†

S greater BDU in last 2 weeks within the omega-3 fatty acids phase

Summary Table 4: RCT evidence of omega-3 fatty acids (more...)

Summary Table 4: RCT evidence of omega-3 fatty acids to influence mediators of inflammation in asthma

	Omega-3 Fatty Acid Arm/Phase		Comparator Arm/Phase
Author, Year, Location	n	Type & Dose/Day	n	Type & Dose/Day	# of S Unique Results Favoring Omega-3 Fatty Acids	Jadad Total Quality/Allocation Concealment (Internal Validity)	Applicability (External Validity)
ADULTS
Arm, 1988, England56,57	NR	3.2 g EPA + 2.2 g DHA	NR	Olive oil (NR)	3/5	4 (Grade: A)/unclear	III
Kirsch, 1988, USA54,55	6	4.0 g EPA ethyl ester (trace DHA)	6	0.1 g EPA EPA ethyl ester (trace DHA)	9/17	3 (Grade: B)/unclear	I
Okamoto, 2000a, Japan66	7	10–20 g perilla seed oil (ALA: NR)	7	10–20 g corn oil	2/2	2 (Grade: C)/ unclear	III
Stenius-Aarniala, 1989, Finland68*	36	20 mL fish oil (EPA+DHA: NR)	36	(1) 20 mL olive oil	2/8	2 (Grade: C)/unclear	III
				(2) 20 mL evening primrose oil
CHILDREN
Hodge, 1998, Australia52	NR	0.72 g EPA 0.48 g DHA + ALA (NR) via canola diet	NR	Omega-6 fatty acids: 1.8 g safflower oil + 1.8 g palm oil + 0.4 g olive oil + sunflower oil diet (NR)	0/1	3 (Grade: B)/unclear	III

N = number of randomized participants; NR = not reported; S = significant;

*

Crossover trial

Summary Table 5: Evidence from other study designs (more...)

Summary Table 5: Evidence from other study designs of omega-3 fatty acids to influence mediators of inflammation in asthma

	Omega-3 Fatty Acid Cohort/Phase		Comparator Cohort/Phase
Author, Year, Location	n	Type & Dose/Day	n	Type & Dose/Day	# of S Unique Results Favoring Omega-3 Fatty Acids	Total Quality (Internal Validity)	Applicability (External Validity)
Ashida, 1997, Japan59†	5	15 g perilla seed oil (ALA: NR)	NA	NA	2/2	3 (Grade: B)	III
Broughton, 1997, USA73†	26	“Low” EPA+ DHA intake: ~0.7 g (mean)	26	“High” EPA+ DHA intake: ~3.3 g (mean)	4/7	3 (Grade: B)	I
Hashimoto, 1997, Japan62†	8	1.8 g EPA	NA	NA	0/2	3 (Grade: B)	III
Masuev, 1997a, Russia60††	27	6.0 g eiconol: EPA+DHA (NR)	7	6.0 g olive oil	1/1	2 (Grade: C)	III
Okamoto, 2000b, Japan71†	26	10–20 g perilla seed oil (ALA: NR)	NA	NA	1/1	4 (Grade: A)	III
Picado, 1988, Spain74†	10	3.0 g EPA+DHA + eucaloric diet	10	3.0 g lactose + eucaloric diet	1/1	5 (Grade: A)	III

n = number of enrolled participants; NR: not reported; NA = not applicable; S = significant;

†

Noncomparative case series;

††

non-RCT

Summary Table 6: RCT evidence of omega-3 fatty acids (more...)

Summary Table 6: RCT evidence of omega-3 fatty acids to prevent asthma in children

	Omega-3 Fatty Acid Group		Comparator Group
Author, Year, Location	n	Type & Dose/Day	n	Type & Dose/Day	# of S Unique Results Favoring Omega-3 Fatty Acids	Jadad Total Quality/Allocation Concealment (Internal Validity)	Applicability (External Validity)
Mihrshahi, 2003, Australia50,51	312	0.8 mg EPA + 3.6 mg DHA per kg body weight (500 mg fish oil) + canola oil/margarine (ALA: NR)	304	500 mg Sunola oil + PUFA oils/margarine	2/10	2 (Grade:C)/adequate	III

n=number of randomized participants; S = significant; PUFA = polyunsaturated fatty acids

Summary Table 7: Observational study evidence of omega-3 (more...)

Summary Table 7: Observational study evidence of omega-3 fatty acids for primary prevention of asthma

Author, Year, Location	Number with asthma	Types of Control (n)	Unadjusted or Adjusted Associations of Dietary Fish Intake and Asthma Prevalence	Study Quality (Internal Validity)	Applicability (External Validity)
ADULTS
Troisi, 1995, USA75	1446	NA	• NS relationship between 6-y risk of asthma and frequency of intake of dark meat fish	3 (Grade: B)	II
			• NS adjusted risk reduction associated with all levels of omega-3 fatty acid intake
ADOLESCENTS
Huang, 2001, Taiwan48,49	36	• Allergic rhinitis (n=115)	• S association between higher frequencies of oily fish intake and asthma prevalence	3 (Grade: B)	III
		• No asthma or rhinitis (n=1,030)
ADOLESCENTS & CHILDREN
Takemura, 2002, Japan76	1673	• Not currently asthmatic (n=22,109)	• S higher asthma prevalence (adjusted) for consumers of 1-2 fish meals/wk than for consumers of 1-2/mo (overall, and only in males)	3 (Grade: B)	III
CHILDREN
Hodge, 1996, Australia77	71	• Normal airways (n=263)	• S lower (unadjusted) risk of asthma in eaters of fresh fish and oily fresh fish	2 (Grade: C)	III
		• Airways hyper-responsiveness only (n=55)	• S lower (adjusted) risk of asthma in consumers of oily fish
		• Wheeze only (n=79)
Satomi, 1994, Japan78	706	• Not asthmatic (n=6,882)	• S negative correlation between asthma prevalence and frequency of fish consumption (e.g., reddish fish)	4 (Grade: A)	III

n = number of evaluated participants; NA = not applicable; S = significant; NS = nonsignificant

Summary Table 8: Studies reporting adverse events, (more...)

Summary Table 8: Studies reporting adverse events, side effects, and counter-indications

	Omega-3 Fatty Acid Arm/Phase		Comparator Arm/Phase
Author, Year, Location	n	Type & Dose/ Day	n	Type & Dose/Day	Safety data
ADULTS
Arm, 1988, England56,57†	NR	3.2 g EPA + 2.2 g DHA	NR	Olive oil (NR)	Withdrew (timing: NR): size & number of capsules not tolerable (n=3; arm: NR); Withdrew after 3 wk (omega-3 fatty acids): hospitalized for acute asthma (n=1)
Broughton, 1997, USA73††	26	‘Low’ EPA + DHA intake: ~0.7 g (mean)	26	‘High’ EPA + DHA intake: ~3.3 g (mean)	Fishy hiccups (omega-3 fatty acids arm: NR); Occasional mild gastrointestinal discomfort (omega-3 fatty acids arm: NR); Exclusion criteria: history of bleeding disorder or delayed clotting time
Emelyanov, 2002, Russia69†	23	200 mg EPA+DHA + 400 mg olive oil	23	600 mg olive oil	Skin itch (1 per study arm); Metallic taste (omega-3 fatty acids arm: 1; control arm: 2)
Kirsch, 1988, USA54,55†	6	4.0 g EPA ethyl ester (trace DHA)	6	0.1 g EPA ethyl ester (trace DHA)	Adverse reactions to aspirin or NSAIDs in high (n=2) & low dose (n=1) arms
Masuev, 1997b, Russia61†††	5	6.0 g EPA+DHA	3	6.0 g Olive oil	Withdrew (timing: NR) due to severe clinical apnea in response to repeated allergen challenge (n=1)
McDonald, 1991, Australia58†*	15	2.7 g EPA + 1.8 g DHA	15	15 g olive oil	Withdrew (timing: NR): problems swallowing capsules (n=2). Exclusion criteria: peptic ulcers, cardiovascular disease, other potential bleeding disorders
Stenius-Aarniala, 1989, Finland68†*	36	20 mL fish oil (EPA+DHA: NR)	36	(1) 20 mL olive oil (2) 20 mL evening primrose oil	Withdrew (timing: NR): could not tolerate taste of oil, or, difficulty keeping diary (n=7; breakdowns: NR) Exclusion criteria: coagulation disorders & diabetes
Thien, 1993, England67†	NR	3.2 g EPA + 2.2 g DHA	NR	Olive oil (NR)	Withdrew in wk 1: nausea & vomiting after taking capsules (omega-3 fatty acids arm: n=1); withdrew after wk 1: size & number of capsules unmanageable (n=6; arm: NR); withdrew in first 2 wk: size & number of capsule unmanageable (n=4; arm: NR); withdrew (timing: NR): difficulty taking capsules & recording data (n=2; arm: NR)
CHILDREN
Hodge, 1998, Australia53†	NR	0.72 g EPA 0.48 g DHA + ALA (NR) via Canola diet	NR	‘Omega-6 fatty acids:’ 1.8g Safflower oil + 1.8g Palm oil + 0.4g Olive oil + Sunflower diet (NR)	Discomfort after taking capsules: omega-3 fatty acids arm (n=1); omega-6 fatty acids arm (n=2)
Nagakura, 2000, Japan64†	15	17.0–26.8 mg/kg EPA; 7.3–11.5 mg/kg DHA (300 mg fish oil)	15	300 mg olive oil	Dropped out: unable to swallow capsules at beginning of study (n=1; arm: control)

n = number of enrolled/randomized participants; NR = not reported; NA = not applicable;

†

RCT;

††

Noncomparative case series

†††

non-RCT

*

Crossover trial

Question 1: What is the evidence for the efficacy of omega-3 fatty acids to improve respiratory outcomes among individuals with asthma?

Qualitative Synthesis of Evidence Regarding Respiratory Outcomes from Study Designs Other Than an RCT

Study characteristics. Nine studies with either a non-RCT or noncomparative case series design, published between 1988 and 2000, were identified as being pertinent to Question 1 (Summary Table 2; Evidence Table 2, Appendix E). Six exclusively involved adults,^{59, 62, 71–74} one enrolled older adolescents and adults 17 to 40 years-of-age,⁶¹ and is considered along with the adult studies, and two evaluated children.^{63, 70}

Only two of the adult studies presented clearly defined inclusion and exclusion criteria,^{62, 73} another adult study provided both, albeit vaguely described,⁷² and four adult studies exclusively described inclusion criteria.^{59, 61, 71, 74} For example, one of the studies excluded adults with a history of bleeding disorders or delayed clotting time.⁷³ Both pediatric investigations failed to report any selection criteria.^{63, 70}

One non-RCT, after identifying adults exhibiting both an acute and late asthmatic response to allergen challenge, divided the eight responders into two groups matched for age, gender, and asthma duration.⁶¹ Four adult studies each employed a single-phase, noncomparative case series design.^{59, 62, 71, 72} Two adult investigations each selected a noncomparative case series, which they then exposed to two interventions in a fixed sequence.^{73, 74} Both pediatric studies used a non-RCT design.^{63, 70} The first one matched the groups for age, diagnosis, and asthma treatment (no data);⁷⁰ the second one did not report how, or if, the groups were matched.⁶³

The nine studies typically involved few participants, with a mean number of 20.3 (range: 5–60).³⁵ A total of 93 children and 90 adults were enrolled. On average, more children than adults were studied (mean: 46.5 participants). Five studies involved no more than ten adults.^{59, 61, 62, 72, 74}

For the eight studies that reported data on study length and intervention length, the average study length was 7.7 (range: 2–14) weeks and the mean intervention length was 6.0 (range: 2–12) weeks. Only four of the eight studies had intervention periods that lasted longer than 8 weeks.^{61–63, 74} Two adult investigations reported the length of a run-in period (each 2 weeks), with neither detailing a protocol.^{62, 74} Neither of the studies employing a two-phase, noncomparative case series design included a washout between their two exposure periods.^{73, 74}

The studies were conducted in six different countries, with Japan represented three times^{59, 62, 71} and Russia twice.^{61, 70} The remaining locations were Poland, Italy, the United States, and Spain. Where data were available, a single site conducted the study. Only two adult studies reported a funding source: a university and industry,⁷³ and a professional society.⁷⁴ Other than the fact that only the pediatric studies exclusively undertook controlled studies, there were no noticeable differences between these and the adult studies with regards to study characteristics.

Population characteristics. Neither pediatric study reported a mean age or a percentage of male participants.^{63, 70} However, they did provide an age range: children aged 1 to 12 years for one study,⁷⁰ and children aged 7 to 17 years for the other one.⁶³ For the six adult studies that reported data on the age of the participants, the average age was 48.0 (range: 17–84) years.^{59, 62, 71–74} The average percentage of males in the five investigations that provided these data was 27.6% (0%-57%).^{59, 62, 71, 72, 74} No authors explicitly stated the racial/ethnic background of the study participants, yet it is likely that Caucasian/Europeans and Asian populations were represented five^{61, 63, 70, 72, 74} and three^{59, 62, 71} times, respectively. Americans constituted the final population, yet its racial/ethnic composition was not described.⁷³

Two adult studies indicated having employed a standard definition of asthma,^{71, 74} one pediatric study provided a vague definition,⁷⁰ and the remaining studies provided no definition at all. For example, Okamoto et al.⁷¹ employed the International Consensus on Diagnosis and Treatment of Asthma criteria, whereas, another study indicated that their participants were in “relative remission,” although this term was not defined.⁶¹ A description of the diagnostic method used to identify asthma was reported on only four occasions, and with varying degrees of detail.^71–74 The one pediatric study enrolling children as young as 1 year-of-age did not provide information regarding how, or if, asthma and wheezing disorders were distinguished.⁷⁰ None of the adult studies that involved the older populations indicated how, or if, asthma and possible COPD were differentiated.^{59, 62, 71, 74}

Four of nine studies indicated the asthma severity of the included study population. This included one study that described subgroups of children with mild or severe asthma,⁶³ one study with adults diagnosed with mild-to-moderate asthma,⁶² and two studies that enrolled adults with mild asthma.^{71, 72} Only one study described the criteria used to classify asthma severity.⁷¹ For four of seven adult studies that reported an asthma duration, the average was 12.9 (range: 3–24.7) years.^{59, 61, 63, 71, 74} One pediatric study reported that the average duration of asthma was 7.36 years for their subgroup with mild asthma, and 9.25 years for their subgroup with moderately-severe asthma.⁶³ None of these studies attempted to define the severity of asthma at baseline, or on-study, in light of how well it was controlled by medication. Of the two pediatric studies with a control group, no information was provided regarding how, or if, the groups were matched on the basis of asthma severity.^{63, 70}

Typically, without clear definition, studies identified their asthma population as having the following concomitant conditions: atopy,^{71, 73} atopic dermatitis,⁷⁰ atopy with allergic asthma,⁶³ atopy with aero-allergens,⁷² hypersensitivity to dust,⁶¹ aspirin-intolerant asthma and nasal polyps (50%),⁷⁴ and cough variant asthma (20%).⁵⁹ One adult investigation evaluated participants with allergic dermatitis (50%) concurrent with hyperlipidemia in its full sample.⁶² Of the two non-RCTs with children, the report information indicated that there was matching across groups for atopic dermatitis,⁷⁰ and to some extent for atopy.⁶³ In the second pediatric study, it was unclear whether, as with the treatment group, the control group contained any children with allergic asthma.⁶³ No information was reported regarding concurrent conditions (or related medications), or the seasons in which the studies took place. One study involved adults hospitalized for asthma.⁵⁹

Regarding the reporting of asthma risk factors, or those with the potential to influence asthma control, very little information was provided. Only one adult study reported having enrolled non-smokers, yet no details were provided regarding their smoking history.⁷³ Data concerning exposure to environmental smoke or a history of early respiratory infections in children were not provided, although one study excluded adult participants if they had had an upper respiratory infection less than 6 weeks prior to the study.⁷³ The two non-RCTs involving children reported no data as to whether their groups were equivalent on any of these or other bases.^{63, 70} Consequently, nothing can be concluded about the influence of these potential confounders on individual study results.

Intervention/exposure characteristics. Two noncomparative case series employed perilla seed oil as their source of omega-3 fatty acids,^{59, 71} four used fish oil,^{63, 70, 73, 74} and three likely used fish oil, given the omega-3 fatty acids that were identified. However, this information was not explicitly stated.^{61, 62, 72} Both pediatric studies employed fish oil.^{63, 70} Only one study described the specific type of fish from which part of their intervention was derived (i.e., sardine).⁷⁴ Four of nine investigations identified the type of omega-3 fatty acid as EPA and DHA,^{61, 72–74} two used EPA exclusively,^{62, 63} two employed ALA,^{59, 71} and one did not report the exact type(s) of omega-3 fatty acids.⁷⁰

The two pediatric non-RCTs compared: fish oil capsules plus a poorly defined hypoallergenic diet with a hypoallergenic diet and no description of the placebo capsules;⁷⁰ and fish oil with sunflower oil, the latter considered a source of omega-6 fatty acids.⁶³ In the first non-RCT, the investigators wished to establish a lower omega-6/omega-3 ratio of fatty acids, yet the data reported were contradictory.⁷⁰ Also, the 4.5 g/day dose was of fish oil, not omega-3 fatty acids. As a result, it was unclear how either exposure was defined in this investigation.

In the two noncomparative case series where the participants received two exposures in sequence, the exposure were: placebo capsules and a eucaloric (32% fat) diet, followed by the same diet, fish oil capsules, and sardine oil in one;⁷⁴ and, a low omega-3 fatty acid intake of fish oil capsules followed by a high omega-3 fatty acid intake in the other.⁷³ In the latter study, fish oil regimens were individualized based on an analysis of prestudy omega-6 fatty acid intake, yielding a low omega-6/omega-3 fatty acid ratio of 1:0.1 in the low fish oil exposure and a high omega-6/omega-3 fatty acid ratio of 1:0.5 in the high fish oil exposure.⁷³ The non-RCT that selected participants on the basis of an acute and late asthmatic response to allergen challenge subsequently assigned participants to receive either EPA and DHA (likely by way of fish oil), or an olive oil control.⁶¹

If greater than or equal to 3 grams per day is considered to be a high daily dose or serving of omega-3 fatty acids, then five studies attained this level: 6 g/day,⁶¹ 3.3 g/day in one noncomparative case series' high-dose phase;⁷³ 3 g/day in the other noncomparative case series' second phase;⁷⁴ 3 g/day in a pediatric non-RCT;⁶³ and, 3 g/day in a noncomparative case series.⁷² One adult noncomparative case series received 1,800 mg/day.⁶² Three studies did not define their omega-3 fatty acid dose or serving.^{59, 70, 71}

The most frequently used method for delivering the fish oil was standardized dosing capsules.^{61, 70, 72, 73} In one study, the method by which participants received the sardine oil was not specified.⁷⁴ In another study, the hypoallergenic diet given along with the fish oil capsules was poorly defined, with no clear indication of the types or amounts of food consumed, or whether consumption varied across the study.⁷⁰ One study described a range for the daily intake of perilla seed supplementation (10–20 g in salad dressing and/or mayonnaise), precluding a precise definition of a serving size or amount of ALA for any participant on any given day.⁷¹ The investigators reported that the adults consumed an average of 14.65 g/day, suggesting variability in intake among study participants.

A second investigation mandated 15 g/day of perilla seed oil consumption, yet there was no report of how, or if, the investigators attempted to control the intake.⁵⁹ One pediatric⁶³ and one adult⁶² study did not describe how their omega-3 fatty acid exposure was delivered. In the reports of the four studies that did not use a completely controlled dosing vehicle, no information was provided to establish a description of the omega-3 fatty acid content.^{59, 70, 71, 74} Thus, for six of the studies employing a design other than an RCT it was impossible to establish the exact definition of the omega-3 fatty acid exposure. In the two pediatric non-RCTs, no data were provided showing that the oil intake was matched, suggesting a likely, between-group difference in caloric intake.^{63, 70} This potential source of confounding was not addressed.

Only one study described the exact timing of the intervention (i.e., at each main meal),⁶³ while a second study mentioned that the intervention was in the morning.⁷³ Few study reports indicated that participants had been instructed to maintain their background diet.^{59, 71} One noncomparative case series was told that they could maintain a free diet (undefined).⁷² Two studies altered the diet of their participants: one to a hypoallergenic diet (undefined),⁷⁰ and one to an eucaloric diet (poorly defined).⁷⁴

No study mandated the intake of omega-6 fatty acids as a cointervention, despite one investigation's attempt to alter the omega-6/omega-3 fatty acid ratio through the consumption of omega-3 fatty acids,⁷³ and another study's attempt to do so through a hypoallergenic diet and fish oil capsules.⁷⁰ The first investigation also excluded adults if they were consuming fish oil supplements or more than one fish meal per week.⁷³

No report indicated the manufacturer of the omega-3 fatty acid intervention although the use of MaxEPA® suggests the involvement of Seven Seas, Ltd.⁷³ Information concerning the purity of the supplementation was never provided.

Cointervention characteristics. There were no data regarding any additional treatment received by the participants in the only study that had a clearly identified concurrent condition (i.e., hyperlipidemia).⁶² There was a dearth of information reported regarding the types and dosing of asthma medication.

One noncomparative case series excluded adults if they were taking any asthma medication.⁷² Another excluded those taking more than 5 mg/day of prednisone or longterm steroids (undefined) started less than one month prior to the study.⁶² A third study asked that no nonsteroidal anti-inflammatory drugs (NSAIDS) be taken,⁷³ and a fourth asked participants to maintain a fixed dose of inhaled corticosteroids and bronchodilator medication during the study.⁷⁴ No other study described whether their participants maintained a constant on-study dose of inhaled or oral corticosteroids. Two studies reported the number of users of each of these drugs.^{59, 71} In one of these, the range of allowable inhaled corticosteroid doses varied greatly across participants (400–1200 ug/day).⁵⁹ Seven of ten adults in one noncomparative case series were corticosteroid dependent, with two taking prednisolone.⁷⁴ However, whether their on-study doses were maintained was not reported. Other asthma medications were also poorly described, such that it was impossible to determine whether on-study doses were kept constant, or whether the types and doses were equivalent across groups in the controlled studies.

Outcome characteristics. The most frequently employed respiratory outcomes were PEF (undefined, yet likely AM),^{61, 63, 72–74} AM PEF,^{59, 62, 71} and FEV₁.^{63, 71, 73} Given the limited descriptions in the individual study reports, it was difficult to determine whether all the pulmonary function tests were based on standard methodologies. The psychometric performance of symptom rating scales was never described. Relative to RCTs, few dropouts or withdrawals were observed.

Summary Matrix 2: Study quality and applicability of (more...)

Summary Matrix 2: Study quality and applicability of evidence regarding respiratory outcomes from study designs other than an RCT

		Quality
		A			B			C
Applicability	I				Author	Year	N
					Broughton	1997	26
	II
		Author	Year	N	Author	Year	N	Author	Year	N
		Masuev	1997b	8	Ashida	1997	5	Gorelova*	1998	33
	III	Okamoto	2000b	26	Hashimoto	1977	8
		Picado	1988	10	Machura	1996	60
		Villani	1998	7

N = number of randomized participants;

*

Pediatric trial

Study quality and applicability. The mean total quality score was 3.6 (range: 2–5), likely indicating good quality. Of note, five of nine studies provided very limited descriptions of study participants.^{59, 60, 62, 70, 73} The quality grades associated with quality scores were entered into the summary matrix. Little variability characterized the applicability rating, with eight of nine studies assigned a level III rating (Summary Matrix 2). This indicates the extremely limited potential for generalization to the typical North American population with asthma.

Four studies exhibited high quality, defined by a total quality score of 4 or 5. However, the generalizability of the results of these four studies to the North American standard set for this review was extremely limited. The only study with strong generalizability potential also exhibited good study quality.⁷³

Question 2: What is the evidence that the possible value (efficacy/association) of omega-3 fatty acids in improving respiratory outcomes is dependent on specific effect modifiers?

Specific effect modifiers identified in consultation with our TEP were candidates for planned investigations with respect to respiratory outcomes. They included the: specific type of fatty acid; specific source; serving size or dose; amount/dose of omega-6 fatty acids given as a co-intervention; ratio of omega-6/omega-3 fatty acids used; fatty acid content of blood lipid biomarkers; absolute fatty acid content of the baseline diet; relative fatty acid content of the baseline diet; tissue ratios of fatty acid (omega-6/omega-3) during the investigative period; intervention length; anti-oxidant use; and, the manufacturer and its product(s).

Question 3: What is the evidence that, in individuals with asthma, omega-3 fatty acids influence mediators of inflammation which are thought to be related to the pathogenesis of asthma?

Qualitative Synthesis of Evidence Regarding Mediators of Inflammation From Study Designs Other Than an RCT

Notable patterns. Six relevant studies published between 1988 and 2000 addressed Question 3 (Summary Table 5; Evidence Table 2: Appendix E). All exclusively involved adults. Five were the noncomparative case series evaluated by Ashida et al.,⁵⁹ Broughton et al.,⁷³ Hashimoto et al.,⁶² Okamoto et al.,⁷¹ and, Picado et al.⁷⁴ The Broughton et al. and Picado et al. noncomparative case series each involved two intervention phases. The sixth relevant study is Masuev's above-noted non-RCT.⁶⁰

A minority of studies reported both inclusion and exclusion criteria.^{62, 73} The six studies typically involved few participants (n = 109), with a mean number of 18.2 (range: 5–34). Three of the studies involved no more than 10 adults.^{59, 62, 74} The studies lasted an average of 7.8 (range: 2–14) weeks and the mean intervention length was 5.5 (range: 2–8.7) weeks. Only one study did not last at least 4 weeks.⁵⁹ Neither of the studies employing a two-phase noncomparative case series included a washout between their exposure periods.^{73, 74}

In the six studies, the average age of participants was 51.2 (range: 19–84) years.^{59, 60, 62, 71, 73, 74} No authors explicitly stated the racial/ethnic background of any of their participants, yet it is likely that Caucasian/Europeans^{60, 74} and Asian populations^{59, 62, 71} were represented twice and three times, respectively. Americans constituted the final population yet its racial/ethnic composition was not reported.⁷³

Only two studies indicated having employed a standard definition of asthma.^{71, 74} Only three reported their diagnostic method.^{71, 73, 74} None of the studies involving the oldest populations indicated how, or if, asthma and possible COPD were differentiated.^{59, 62, 71, 74} Only one of the studies described criteria classifying asthma severity.⁷¹ None of these studies attempted to define the severity of asthma at baseline, or on-study, on the basis of how well it was controlled by medication. In the non-RCT, no information indicated whether the groups had been matched on the basis of asthma severity.⁶⁰

Conditions concomitant to asthma were poorly defined. One of the studies evaluated participants with allergic dermatitis (50%) concurrent with hyperlipidemia in its full sample.⁶² No information was reported regarding other concurrent conditions (or related medications) or the seasons within which the studies took place. One study involved adults hospitalized for asthma.⁵⁹ Regarding the reporting of asthma risk factors, or those with the potential to influence asthma control, very little information was provided. Only one study reported having enrolled non-smokers yet no details were provided regarding their smoking history.⁷³ Consequently, nothing can be concluded about the influence of these potential confounders on individual study results.

Various sources of omega-3 fatty acids were used. If a high daily dose or serving of omega-3 fatty acids for adults is defined as greater than or equal to 3 g of omega-3 fatty acids, then three studies met this criterion.^{60, 73, 74} Two studies did not define their omega-3 fatty acid dose or serving.^{59, 71} One study described a range of intake for the daily use of perilla seed oil-supplemented salad dressing and/or mayonnaise (10–20 g), precluding a precise definition of a serving size for any participant on any given day.⁷¹ The investigators reported that the adults consumed only 14.65 g/day, likely suggesting variability in intake among study participants. A second investigation mandated 15 g/day of perilla seed oil consumption yet there was no report of how, or whether, the investigators attempted to control the intake.⁵⁹ One study did not describe how their omega-3 fatty acid exposure was delivered.⁶² In the studies that did not use a completely controlled dosing vehicle, no information was provided to establish a description of the omega-3 fatty acid content.^{59, 71, 74} For four studies it was thus impossible to establish exactly the definition of the omega-3 fatty acid exposure.

Two studies indicated that participants were told to maintain their background diet.^{59, 71} One study altered the diet of their participants to an eucaloric diet (poorly defined).⁷⁴ No study mandated the intake of omega-6 fatty acids as a cointervention, although one study did attempt to alter the omega-6/omega-3 fatty acid ratio through the consumption of omega-3 fatty acids.⁷³ Information concerning the purity of the supplementation was never provided.

There are no data concerning the treatment received by participants in the study in which participants were reported to have a clearly identified concurrent condition (i.e., hyperlipidemia).⁶² There is a dearth of information reported regarding the types and dosing of asthma medication. One noncomparative case series excluded individuals taking more than 5 mg/day of prednisone, or longterm steroids (undefined) that were started less than one month prior to the study.⁶² A second study requested that no NSAIDS be taken,⁷³ and a third asked participants to maintain a fixed dose of inhaled corticosteroids and bronchodilator medication during the study.⁷⁴ No other study described whether their participants maintained a constant on-study dose of inhaled or oral corticosteroids. A few studies reported the number of users of each of these drugs;^{59, 71} in the one study,⁵⁹ the range of allowable inhaled corticosteroid doses varied greatly across participants (400–1200 ug/day). Seven of ten adults in one noncomparative case series were steroid-dependent, with two of the adults taking prednisolone;⁷⁴ however, whether their on-study doses were maintained was not reported. Other asthma medications were also poorly described and hence it was impossible to determine whether on-study doses were kept constant, or, whether the types and doses were equivalent across groups in the only controlled study.

Outcome characteristics. The most frequently employed intermediate outcomes were the various leukotriene series. Given the limited descriptions in the individual study reports, it was difficult to determine whether all the pulmonary function tests were based on standard methodologies.

Summary Matrix 4: Study quality and applicability of (more...)

Summary Matrix 4: Study quality and applicability of evidence regarding mediators of inflammation from study designs other than an RCT

		Quality
		A			B			C
Applicability	I				Author	Year	N
					Broughton	1997	26
	II
		Author	Year	N	Author	Year	N	Author	Year	N
	III	Okamoto	2000b	26	Ashida	1977	5	Masuev	1997a	34
		Picado	1988	10	Hashimoto	1977	8

N = number of randomized participants

Study quality and applicability. The mean total quality score was 3.3 (range: 2–5), likely indicating good quality. Of note, two of six studies provided very limited descriptions of study participants lost to followup.^{60, 73} The quality grades associated with quality scores were entered into the summary matrix. Little variability characterized the applicability rating, with five of six studies assigned a level III rating (Summary Matrix 4). This indicates the extremely limited potential for generalization to the typical North American population with asthma.

Two noncomparative case series exhibited high quality, defined by a total quality score of 4 or 5.^{71, 74} However, the generalizability of the results of these studies to the North American standard set for this review was extremely limited. The only study with strong generalizability potential also exhibited good study quality.⁷³

Question 4: Are omega-3 fatty acids effective in the primary prevention of asthma?

Qualitative Synthesis of Observational Study Evidence Regarding Primary Prevention

Study characteristics. Adult and pediatric studies are described separately. All studies were published between 1994 and 2003 (Summary Table 7; Evidence Table 4: Appendix E). The study evaluating both young children and adolescents is included with the pediatric investigations.

All but one study reported both inclusion and exclusion criteria, with one pediatric study failing to explicitly state exclusion criteria.⁷⁸ Two studies provided very few details regarding their sampling procedure.^{76, 78} The full sample sizes varied, ranging from a cross-section of 808 school-age children⁷⁷ to 121,700 nurses in the Nurses Health Study.⁷⁵ Two studies were funded by government,^{48, 77}, one by a medical association,⁷⁶ one by the NIH,⁷⁵ and one did not report a funding source.⁷⁸

Population characteristics. The adult study followed a cohort of exclusively female nurses aged between 34 and 68 years.⁷⁵ Approximately half of the participants in the other studies were male. Of the two studies that included adolescents, one examined individuals between the ages of 6 and 15 (mean: 10.41) years,⁷⁶ and the ages of the participants in the other study ranged between 3 and 17 (mean: 14.7) years.⁴⁸ Children in the two pediatric investigations ranged in age between 6 and 11 (mean: NR) years,⁷⁸ and 8 and 11 (mean: 9.5) years.⁷⁷ While it could be surmised that the three studies that involved Asian populations,^{48, 76, 78} the Australian project⁷⁷ likely included those of primarily Caucasian/European descent, whereas, the American study provided no race/ethnicity details.⁷⁵

Each study evaluating asthma prevalence included a subset of asthmatic participants. The only study to assess incidence of asthma excluded all asthmatic individuals prior to following a cohort of nurses prospectively.⁷⁵ Of those including asthmatic participants, the asthma sample sizes ranged from 36 adolescents⁴⁸ to 1,673 children and adolescents.⁷⁶ Three of the latter four samples of asthmatics^76–78 included an average of 60.2% males, exceeding the 50% value for their full sample populations. Methods to identify asthma varied, and included questionnaires asking about a physician-assigned diagnosis of asthma,^{48, 75, 78} a modified American Thoracic Society questionnaire,⁷⁶and testing combined with a clinical assessment.⁷⁷ Asthma severity data were not reported for any study. Only the report of the Nurses Health Study described the range of asthma medications used by the participants.⁷⁵ These included various types of corticosteroid.

The four pediatric studies each defined a control sub-sample of individuals without asthma,^{48, 76–78} with their sizes ranging from 263 children⁷⁷ to 22,109 children and adolescents.⁷⁶ Only two reports described whether, and how, its sub-populations varied. Hodge et al. found more children with atopy in their asthmatic group than in their other subpopulations; in addition they found no difference between groups with respect to a history of early respiratory infections or parental history of asthma.⁷⁷ Takemura et al. reported that, relative to non-asthmatics, the asthmatic children and adolescents were younger, more likely to be male, and more likely to have parents with a history of asthma.⁷⁶

Exposure characteristics. The exposures of interest to the present systematic review involved the dietary intake of various types of fish, indicating the possible presence of EPA and DHA. However, the specific amounts (in grams) of these omega-3 fatty acids were never reported. Four studies reported the types of fish, arguing that certain fish (reddish;⁷⁸ oily;^{48, 77} dark meat⁷⁵) contained greater amounts of omega-3 fatty acids. One study did not report the types of fish.⁷⁶ The timeframe of food intake varied between respondents but was reported to be within: the past year;^{75, 77} the current diet;^{76, 78} and, both the last month, and, 24-hour recall.⁴⁸ Every study employed at least a semi-quantitative questionnaire with which to collect data, although one also included interviews with the participating adolescents.⁴⁸ The various questionnaires provided different response options (e.g., “never” to “daily;”⁷⁷ “less than once a month” to “more than 4–5 times a week;”⁷⁸ “almost never” to “at least 3–4 times a week”).⁷⁶ Assessments involving children required that parents provide the data.^{48, 76–78} While investigators were invariably focused on fresh fish consumption (vs canned products), no studies factored into their analysis the ways in which the fish were prepared or the impact of the preparation method (e.g., frying) on the omega-3 fatty acid content.

Outcome characteristics. The key outcomes were the prevalence^{48, 76–78} and incidence⁷⁵ of asthma and its core symptoms.

Summary Matrix 5: Study quality and applicability of (more...)

Summary Matrix 5: Study quality and applicability of observational study evidence regarding primary prevention

		Quality
		A			B			C
Applicability	I
	II				Author	Year	N
					Troisi*	1995	1446
		Author	Year	N	Author	Year	N	Author	Year	N
	III	Satomi	1994	706	Huang	2001a	36	Hodge	1996	71
					Takemura	2002	1673

N = number of randomized participants

*

Adults (nurses) only

Study quality and applicability. The mean total quality score was 3.2 (range: 2–4), likely indicating good quality. All five studies failed to describe their exposures adequately, and two failed to satisfactorily describe their study participants.^{76, 77} The quality grades associated with quality scores were entered into the summary matrix. Virtually no variability characterized the applicability rating, with four of five studies assigned a level III rating (Summary Matrix 5). This indicates the extremely limited potential for generalization to the typical healthy North American population, or to those at risk for asthma.

One observational study exhibited high quality, defined by a total quality score of 4.⁷⁸ However, its applicability was extremely limited. The only study with reasonable applicability also exhibited good quality.⁷⁵

Question 5: Among Individuals with Asthma, do Omega-3 Fatty Acids Alter the Progression of Asthma (i.e., Secondary Prevention)?

There were no studies found that investigated this question.

Question 6: What is the Evidence for Adverse Events, Side Effects, or Counter-Indications Associated with Omega-3 Fatty Acid Use to Treat or Prevent Asthma (DHA, EPA, DPA, ALA, Fish Oil, Fish)?

Question 7: What is the Evidence that Omega-3 Fatty Acids are Associated with Adverse Events in Specific Subpopulations of Asthmatic Individual such as Diabetics?

There were no studies found that investigated this question.

Overview

Twenty-six studies investigated five of the seven questions addressed in this systematic review of the evidence concerning the health effects of omega-3 fatty acids in asthma. The question of secondary prevention and the question of safety related to omega-3 fatty acid use in subpopulations of asthmatic could not be addressed since there were no studies addressing either of these topics. Eleven RCTs (ten treatment; one primary prevention) and 15 studies employing other designs (ten treatment; five primary prevention) were included. Three of the former and six of the latter exclusively involved either children or adolescents. It is likely that, other than Ashida et al.'s noncomparative case series lasting 2 weeks,⁵⁹ all treatment studies lasted long enough to demonstrate that a difference could be found in terms of respiratory outcomes and mediators of inflammation. Relevant studies could only be synthesized qualitatively according to the question(s) that they addressed. Reasons for choosing to forego meta-analysis follow from a critical analysis of the evidence base. After the response to each research question is presented, the discussion turns to the broader implications and issues raised by the findings.

The Evidence

There is not enough consistent evidence, or sufficient evidence from methodologically sound and adequately powered studies with which to conclude definitively that omega-3 fatty acids are or are not efficacious in improving respiratory outcomes in adults or children (Question 1). Failure to control for confounding in over half the RCTs also made it difficult, if not inappropriate, to summarize their data. The greater inability to control for confounders in study designs less “naturally” rigorous than RCTs complicated the synthesis of those data.

In light of the available information, the inconsistency among study results may be attributable to the heterogeneity in definitions of the: settings (e.g., hospital vs outpatient; countries); populations (e.g., age; gender; clinical picture of asthma, including its severity and concomitants, or triggers with the potential to impact asthma control); interventions and their contrasts with comparators (e.g., different types and amounts of oil and omega-3 fatty acid contents; controlled vs uncontrolled dosing), and cointerventions (e.g., asthma medication with varying capacities to control asthma in the short-term or longterm). This observation applies to all patterns of results relating to Questions 1, 2, 3, and 4. Explicitly defined study quality was good for the various types of study design, with the prominent limitation for RCTs being limited blinding, and the key limitation for studies using other designs being the poor description of study participants. All but one RCT⁵⁴ and one two-phase, noncomparative case series ⁷³ exhibited very restricted applicability to the typical North American population of asthmatics.

Adult RCTs revealed a somewhat contradictory picture of efficacy with respect to this systematic review's primary outcome, FEV₁. One very small adult study (n = 14) that employed uncontrolled dosing of perilla seed oil and corn oil (control) over a short intervention period (n = 4 weeks) reported a significant effect. However, two RCTs each observed no benefit relating to an omega-3 fatty acids intervention. One compared high and low doses of EPA ethyl ester⁵⁴ over 16 weeks in a small study (n = 12), whereas the second investigated the benefit of low dose EPA/DHA (vs. olive oil) over ten weeks in the systematic review's highest quality RCT.⁶⁹ The latter involved one of the largest sample populations (n = 46) included in the evidence review. Emelyanov et al. also demonstrated good control of three confounding factors (see Critical Analysis) while providing one of the most rigorous methods to select its asthma population.⁶⁹ No studies of adults in a non-RCT or noncomparative case series investigated this outcome. With regard to studies of children, one RCT⁵² and a non-RCT⁶³ observed no benefit in terms of FEV₁. Thus, while it might be tempting to conclude an absence of efficacy based on the solid Emelyanov et al. study,⁶⁹ the fact that there were few studies to consider makes the most balanced understanding one that suggests more research is needed before anything definitive can be concluded about the impact of omega-3 fatty acids on FEV₁. Moreover, a change in this outcome perhaps should not have been expected in the Emelyanov et al. trial because they utilized a low dose of omega-3 fatty acids, as well as randomized mild asthmatics. That said, exactly to what the other observed clinical effects (e.g., AM PEF) in their study may be attributed (e.g., another component in Lyprinol) is unclear.

With respect to several other respiratory outcomes, there is likewise no unequivocal response to the question of efficacy for adults or children. For AM PEF in adults, two RCTs,^{66, 69} including Emelyanov et al.,⁶⁹ and three noncomparative case series^{59, 62} showed a significant benefit for omega-3 fatty acid supplementation. However, four RCTs^{57, 58, 67, 68} and a noncomparative case series⁷² reported no benefit. The pediatric non-RCT also revealed no benefit.⁶³

For PM PEF in adults, five RCTs including Emelyanov et al.,⁶⁹ showed no benefit whereas two noncomparative case series noted a significant effect.^{59, 62} While the RCT evidence likely deserves to be “weighted” more heavily, and suggests no efficacy, this pattern requires interpretation within the larger context of the various outcomes' findings. Based on two RCTs, ^{57, 67} no effect was observed for diurnal PEF variability in adults. Pediatric studies did not employ either of these latter two outcomes.

Results varied with respect to adult bronchodilator use. Three RCTs^{57, 58, 67} reported no benefit (i.e., decreased use) associated with omega-3 fatty acid supplementation. One noncomparative case series with two exposures observed a significant increase in bronchodilator use associated with deteriorating airflow obstruction in the last two weeks of fish oil supplementation (versus placebo).⁷⁴ Emelyanov et al.'s adult RCT,⁶⁹ and a non-RCT with children,⁷⁰ each observed a benefit associated with omega-3 fatty acid use. The latter enrolled children as young as one year of age, however.

Observing results relating to subjective ratings of respiratory function, including asthma symptom scores and severity scores, with each based on widely varying definitions and informants (i.e., participants; professionals; parents), revealed a significantly greater decrease in daytime wheeze for adults receiving omega-3 fatty acids in one RCT,⁶⁹ and a significant decrease in symptom and asthma scores in two noncomparative case series.^{59, 62} Two adult RCTs^{57, 67} noted no benefit with respect to airways responsiveness to histamine challenge, whereas one adult RCT⁵⁷ and a non-RCT of adults⁶¹ each reported some value associated with omega-3 fatty acid use in blunting late airways responses to allergen challenge. The remaining observations from both RCTs and other designs involved one study per outcome, including poorly defined subjective ones (e.g., “daily life score”) and several others reflecting pulmonary function (e.g., FVC). No discernible patterns were observed supporting an unequivocal interpretation of omega-3 fatty acids' efficacy.

Given the largely inconsistent picture of efficacy within and across respiratory outcomes, it is impossible to conclude one way or the other that omega-3 fatty acids are an efficacious adjuvant or monotherapy in improving respiratory outcomes in adults or children. This view is perhaps best illustrated by what was observed with respect to the primary outcome, FEV₁. In general, very few studies were available with which to address the question for adults, and even fewer for children and adolescents. Even though study quality, as operationally defined in the present review, was not an obvious shortcoming of the 20 included treatment studies, the very limited generalizability potential for all but two of them may be taken to suggest that answering Question 1 requires more research conducted with North American samples. Additional observations are highlighted in subsequent sections.

The observations derived from the indirect assessment of the possible influence of predefined or study-defined covariates (Question 2) on the results of treatment studies are highly unreliable. With almost no direct tests of the predictive utility of effect modifiers, or the possibility of subgroup meta-analysis (see below), and with few studies consistently observing significant effects for a given outcome, the evaluation yielded few clear observations. Nevertheless, it did highlight one exposure potentially worth exploring in future empirical investigations of the health effects of omega-3 fatty acids in asthma.

It was noted that perilla seed oil supplementation, provided in an uncontrolled fashion to adults across various studies, exclusively produced significant clinical effects in favor of this omega-3 fatty acids exposure (12/12). This observation is based on results from a small RCT (e.g., FEV₁, AM PEF)⁶⁶ and two noncomparative case series (e.g., AM PEF; PM PEF).^{59, 71} However, one issue requiring resolution is the amount of omega-3 fatty acid content participants actually received from this supplementation. Aside from the perilla seed oil observation, too many tenuous connections were observed to permit their extrapolation. The relationship of variables such as the type, source, or dose of the omega-3 fatty acids could not be investigated with any precision. Nevertheless it seemed to be the case that none of the specific definitions, or levels, of the predefined or study-defined covariates was associated exclusively with a significant effect (e.g., high or low dose). Without the option of meta-analysis, it is difficult to respond adequately to Question 2. It must be concluded that, at present, effect modifiers responsible for any significant asthma-related benefits accruing to omega-3 fatty acids supplementation cannot be identified. This exploration was complicated by the fact that few significant effects were found.

It is likewise unfeasible to conclude one way or the other that omega-3 fatty acids positively influence the lipid mediators of inflammation in adult studies in accordance with the biological model implicating the lipoxygenase and cyclooxygenase pathways in asthma. The effects with respect to any of leukotriene series when omega-3 fatty acids were given were not consistently observed across this collection of studies. Possible reasons include poorly designed studies, varying populations, small sample sizes, and, varying or problematic laboratory methods. Moreover, virtually no mediators of inflammation other than the lipid variety were found to have been investigated (e.g., TNF-a). With the omega-3 fatty acids exposure, there was a significantly suppressed generation of LTB₄ by various types of leukocyte observed in two small RCTs totaling 16 participants^{54, 66} and an even smaller noncomparative case series, (n = 5)⁵⁹ but no effect was observed in a larger RCT (n = 25).⁵⁷ An inconsistent picture was observed with respect to the increased generation of LTB₅ by various types of leukocyte, with a significant increase revealed by one small RCT (n = 12)⁵⁴ and a null effect found in a larger RCT (n = 25).⁵⁷ One RCT noted the significant suppression of total LTB compounds by leukocytes.⁵⁷ Yet, one small RCT (n = 14)⁶⁶ and two noncomparative case series^{59, 71} totaling 31 adults consistently found a significantly suppressed generation of LTC₄ by peripheral leukocytes. In these three studies, perilla seed oil had been delivered via uncontrolled servings.

The nonsignificant generation of PGE by various leukocytes in one small RCT (n = 12)⁵⁴ is contrasted with the significantly higher plasma PGE₂ and significantly lower plasma PGF_2-alpha reported for omega-3 fatty acid supplementation in a crossover RCT (n = 36), the latter a study that failed to provide a washout in comparing uncontrolled servings of fish oil supplementation, olive oil, and evening primrose oil.⁶⁸ This same crossover study also revealed nonsignificant effects with respect to plasma TxB₂, plasma 6-keto-PGF_1-alpha, in addition to urinary PGE₂, PGF_2-alpha, TxB₂, and 6-keto-PGF_1-alpha. Yet, Picado's small (n = 10), two-exposure noncomparative case series reported a significant decrease in TxB₂ associated with omega-3 fatty acids use.⁷⁴ In their noncomparative case series (n = 26), Broughton et al. found that significantly lower urinary LTE₄ excretion and significantly higher urinary LTE₅ excretion were associated with high, but not low, dose EPA.⁷³ Hashimoto et al.'s noncomparative case series noted a nonsignificant change in urinary LTB₄ and LTE₄ excretion.⁶²

Two RCTs found a significant suppression of polymorphonuclear leukocyte chemotaxis in response to various stimuli (C5a, LTB₄, and two fMLP concentrations) in a total of 37 adult with asthma, ^{54, 57} while the smaller of the two RCTs (n = 12) also reported a nonsignificant suppression of mononuclear leukocyte chemotaxis to various stimuli (C5a, LTB₄) associated with omega-3 fatty acid use.⁵⁴ Finally, Hodge et al. reported a nonsignificant change in TNF-a production in their pediatric RCT that compared omega-3 fatty acid with omega-6 fatty acid supplementation.⁵²

The only consistent impacts of omega-3 fatty acids on mediators of inflammation involved the suppression of LTC₄ and of polymorphonuclear leukocyte chemotaxis to various stimuli. However, all of the results must be interpreted with caution given the small sample sizes and the fact that the findings of significant effects for the same outcome involved different intervention-comparator contrasts, as well as varying doses of omega-3 fatty acids. As with the evidence regarding Question 1, considerable clinical heterogeneity characterizes these studies. Their average study quality was good, and, their applicability was very restricted. The implications of these observations are described later.

Dietary fish consumption, including oily fish intake, was assessed primarily through retrospective food-frequency questionnaires, and appeared to serve as a primary prevention for asthma in two pediatric populations (Question 4).^{77, 78} However, asthma prevalence and fish, or oily fish, intake were significantly and positively related in studies that included adolescents from Asia,^{48, 76} with one of these studies also including some children.⁷⁶ In a prospective study that followed nurses, no association was found between adult-onset asthma and dietary fish intake.⁷⁵

One possible interpretation of the inverse relationship between age and the ability of regular fish intake to protect against asthma is that, the sooner people, especially children at risk, are exposed to omega-3 fatty acids, the more likely they will be protected. It is possible that even low fish (oil) intake has effects on specific immunological factors inherent to the development of asthma in childhood, which may no longer be modifiable later in life.⁷⁹ Early in life, omega-3 fatty acids may reduce inflammatory responses to allergens.⁵² Eventually, a critical period may be identified.

Mihrshahi et al.'s factorial RCT is, in large part, a study evaluating the impact of an omega-3 fatty acid regimen (vs placebo), initiated prebirth, on neonates at risk for asthma, given that at least one parent or sibling had received this diagnosis.⁵¹ Their interim results indicated a limited benefit accruing to the omega-3 fatty acid exposure, yet 18 months is likely too early in life to reliably identify asthma. Final followup at five years of age should provide a clearer picture of the value of omega-3 fatty acids as primary prevention. Regarding the prevention studies, study quality was better, on average, for the observational studies than for the single RCT; and, as was the case for treatment studies, almost no studies even remotely resembled the North American population standard established in this review. Problems with these prevention studies are enumerated below, including speculation as to why the protective effect was not observed in the studies enrolling adolescents.

No safety profile relating to omega-3 fatty acids as an exposure was reported for primary prevention studies (Question 6); evidence from the remaining studies suggests that the safety profile in treatment studies was good. Most of the adverse events were related to the capsule delivery of oils, rather than to the oils per se.^{57, 58, 64, 67} On several occasions, an inability to swallow capsules led to withdrawal from a study. Other participants may have had difficulties taking eighteen capsules a day of oil in two specific RCTs, yet these difficulties were not reported.^{57, 67} The one moderately serious reaction observed was an undefined number of episodes of nausea and vomiting after ingesting the fish oil capsules, and this led to a withdrawal.⁶⁷ Unspecified numbers of children and adults experienced some mild gastrointestinal discomfort, but not all individuals had been receiving the omega-3 fatty acid exposure.⁵² Fishy hiccups or burping were a rare complaint. By far the most serious event linked to a treatment study involved severe apnea associated with repeated allergen challenge, not the omega-3 fatty acid exposure.⁶¹

Thus, either omega-3 fatty acid supplementation in these studies did not constitute a notable safety problem, or safety data were under-reported. Given what has been observed by others,³⁵ the first interpretation appears to be, at the very least, tenable. While the U.S. Food and Drug Administration has in the past recommended three grams as the maximum daily intake of EPA and DHA, even minor safety concerns associated with larger doses were rare in the present collection of studies. A critical analysis puts into perspective the less than definitive answers to the investigated research questions.

Critical Analysis

Many limitations and problems characterize the present evidence base. To begin with, for the treatment studies, only a small minority of RCTs^{52, 54, 69} and studies using other designs^{62, 72, 73} reported both inclusion and exclusion criteria. Most studies were very small, with many of the RCTs likely being underpowered. Relative to the RCTs, studies with designs exhibiting considerably less potential rigor were even smaller, shorter in duration, and provided less information. Thus, most of the following critique focuses on the RCTs. It must also be recalled that only two treatment studies and one investigating primary prevention demonstrated good and somewhat restricted applicability, respectively. The results of all other studies likely cannot be generalized to the North American reference standard defined in the present review.

The majority of studies were poorly reported, often failing to include key details that could clearly identify the target population, the intervention/exposure it received, or any other treatments that they may have been receiving for asthma. For example, some of the studies did not present any, complete, or non-contradictory, population information as basic as demographic details (e.g., age; gender distribution), diagnosis-related information beyond a label of “asthma” (e.g., definition, including severity, duration, concomitants/triggers with the potential to influence asthma control; method of diagnosis), or lifestyle/racial/ethnic factors with the potential to influence asthma or the effectiveness of its treatment (e.g., background diet).^{65, 66} Yet, three of 20 treatment studies did acknowledge that their asthma participants had been in-patients in controlled, hospital environments.^{59, 64, 66} Investigators of a noncomparative case series claimed their asthma participants were in “remission” yet, while they never defined the term, they continued to describe them as if they were currently asthmatic.⁷² Some RCT reports stated how many adults or children had been randomized, but did not report numbers of participant per study arm.^{52, 57, 65, 67} One RCT reported data and information only for study completers,⁶⁷ and the scarcity of information contained in all reports of the treatment studies suggest that this was not the only instance. One pediatric RCT⁶⁴ included children under the age of five, whereas a non-RCT enrolled children as young as one year of age.⁷⁰ In neither of these reports did the authors assure, with or without data, that they had adequately distinguished between wheezing disorders and asthma. Early transient wheeze is not a reliable predictor of asthma.⁵¹ Some studies involving older adults, with some current or ex-smokers, did not report if they had ruled out COPD.^{66, 68} Almost no information was made available regarding concurrent conditions (e.g., hyperlipidemia) or their treatment.⁶² The latter could have interacted with the participants' asthma treatment.

The same limitations are observed with respect to the reporting of characteristics defining the intervention/exposure (e.g., amounts of omega-3 fatty acid), comparators (e.g., “placebo”), and the allowable or mandated types and doses of cointervention (e.g., types and doses of asthma medication). Some authors defined an intervention/exposure only in terms of the amount of oil known to contain omega-3 fatty acids, while others solely described the amount of omega-3 fatty acid content without any indication of the amount of oil consumed.^{59, 66, 68, 71} When whole foods were a component of the on-study intervention/exposure (e.g., hypoallergenic diet), little descriptive information was reported.^{52, 64, 70} Problems relating to the failed control of two confounders (i.e., uncontrolled serving/dose sizes; varying uses of asthma medication with the potential to influence asthma) are described below. Finally, outcomes involving subjective measures of respiratory function tended to be poorly defined, if at all.

What data were reported suggest the presence of flawed methodologies and designs. This observation contradicts the picture of good study quality (i.e., internal validity) for both RCTs and other designs obtained through formal quality assessment. An implication of this discord is addressed later. Nevertheless, restricted, failed or no attempts at blinding, was one of the biggest threats to the internal validity of RCTs. It was the Jadad quality component for which the most studies failed to receive a single point (n = 2/10);^{66, 68} and, in the case of the Stenius-Aarniala trial, the authors also suggested that their prestudy familiarity with participants likely influenced the latter's improvements in functioning.⁶⁸ Very few studies provided information regarding their run-in protocol or duration. Two crossover trials and two noncomparative case series, each of the latter with two exposures, did not include a washout period.^{58, 68, 73, 74} In these uncontrolled investigations, participants always received the exposures in the same order.^{73, 74} The timing of the delivery of the exposures was seldom described.^{57, 69} Problems associated with some of the delivery methods were presented in the results section regarding safety (e.g., 18 large capsules/day).^{57, 67} Having so many participants drop out because of such a difficulty raises questions about the levels of compliance in those who remained in the studies. Also, few treatment studies framed their results in terms of whether they took place during pollen season, a time of great instability in pulmonary functioning for many children and adults.

Very few studies reported having analyzed their data on an intention-to-treat basis.^{58, 64, 69} Some studies reported no withdrawals or dropouts, yet did not indicate that they had analyzed their data in accordance with this principle.⁶⁵ Moreover, most treatment studies did not report the results of, or possibly never undertook, tests of significance evaluating between-arm differences in outcomes. They did not present or possibly analyze their data in terms of between-group differences in (percent) change from baseline in a particular outcome. Rather, they tended to present results of data analyzed separately from each group. While this may be appropriate to assess whether, for example, EPA levels in cell membranes actually increased in an omega-3 fatty acids intervention group, independently analyzing study groups' data means failing to benefit from the RCT's capacity to control for the effects of certain key factors that could, in less rigorous designs, make difficult or impossible the relatively unequivocal interpretation of between-group results. The number of significant results reported by studies may be exaggerated as a result.

One option entertained was to have the present review team derive effect sizes and confidence intervals for each study's respiratory outcome data. However, considerable data were not reported in trials, making it difficult to calculate these estimates and their precision. For example, estimation of the effect size requires an estimate of the standard deviation of the difference in outcomes between the treatment and control groups. When the outcome is measured as (percent) change from baseline, incomplete reporting of results may complicate or prevent estimation of this standard deviation. When pretest and posttest means and standard deviations are given, as was the case regarding many of the included trials' outcomes, but standard deviations of the change (from pretest to posttest) are not, one possibility is to use a variance imputation strategy such as that proposed by Follmann et al.⁸⁰

Since observations on individuals are generally correlated, the standard deviation of the change within an individual depends on this correlation. Follmann et al. note that if the population measurement variances are equal pretest and posttest, then “presumably the correlation is at least 0.5, otherwise the pretest-posttest design is less efficient than a test based on just the final measurements.” Assuming a correlation of 0.5 (or a correlation estimated from similar trials with more complete reporting of results) then leads to an imputed estimate of the standard deviations of the change. A further assumption that the population change variances are equal in the treatment and control groups then leads to an estimate of the standard deviation of the difference in outcomes between the treatment and control groups. Although this approach may be intuitively appealing, it lacks empirical verification and depends on assumptions of equality of measurement and change variances that may be difficult to verify. Standard deviations imputed in this manner may be very sensitive to the assumed correlation, and resulting effect size estimates may be biased. As a result, it was decided not to follow this strategy. Instead, available results were entered into the respective trials' evidence tables.

That so much important information was missing from reports concerning these next variables' assessment/status and significance, or because available descriptions clearly indicated failures to adequately deal with them, suggest that in planning their studies most investigators did not appreciate the need to control for the threat to the internal validity of their treatment studies posed by at least three key confounders (i.e., population, intervention/exposure, cointervention). If uncontrolled for in treatment studies of any design, these could complicate or prevent the meaningful interpretability of results regarding the utility of omega-3 fatty acids for asthma. That is, failure to control for their possible influence could make it difficult to unequivocally attribute any result (e.g., significant or null) to the actions of this intervention/exposure. Poor reporting of study details further complicates matters by making it difficult to rule out the possibility that these factors alone or in combination could account for the study results, perhaps as well as the omega-3 fatty acids exposure could. How the following observations relate to the issue of study quality is discussed later.

Asthma is a multi-factorial phenomenon that can be triggered by numerous stimuli or circumstances, including exposure to feathered or furry pets, respiratory infections, smoking, or exposure to secondhand smoke and other pollutants; and, asthma control afforded by medication can deteriorate when an asthmatic is exposed to these. It is thus important in studies evaluating the clinical benefit of an asthma intervention that factors with the potential to influence asthma control be assessed and taken into consideration in trying to understand the results.

For example, if during pollen season a greater number of children seriously affected by pollen are randomized to the placebo arm than to the omega-3 fatty acids intervention arm, then a greater frequency and severity of exacerbations indicating a loss of asthma control in the control group could influence the picture of omega-3 fatty acids' efficacy, when expressed as a between-arm difference in objective or subjective measures of respiratory function. A significant treatment effect might have to be attributed to both the benefit from taking omega-3 fatty acids in the active treatment arm and the significant loss of asthma control in the control group. That said, rarely did a study provide clear information as to whether it had been conducted in or out of pollen season, a time when many asthma reactions are triggered.⁶⁷ This raises the possibility that this factor was not adequately controlled for. Only one RCT noted that their study took place out of pollen season,⁵⁷ while another intentionally assessed participants both during and outside pollen season.⁶⁷

In general, studies did a poor job of describing how, and if, factors with the potential to influence asthma control were handled. Whether atopic participants, or those with more severe forms of asthma, were distributed equally across study arms was rarely reported.⁶⁶ The few RCTs that did report on these factors also demonstrated that randomization does not necessarily neutralize possible key confounding influences by equally distributing participants characterized by these factors across study arms. For example, one RCT noted that the severity ratings were higher at baseline in the omega-3 fatty acids group, which may have contributed to a significant decrease in these scores across treatment.⁶⁴ The same pediatric RCT noted, without data, significant between-arm differences in the amount of on-study asthma medication required for acute asthma attacks. Yet, such reports of possible confounders were rare.

Thirty percent of each of the included RCTs and studies with other designs mandated that participants consume certain servings of oil or food rich in omega-3 fatty acids without the use of standardized “dosing” (e.g., capsules). One RCT⁶⁶ and two noncomparative case series^{59, 71} each employed perilla seed oil supplementation, while another RCT delivered its fish and control oils by having it poured from masked bottles.⁶⁸ On several occasions, an intake range was specified (e.g., 10–20 g/day) for delivery by spoon⁶⁶ or by being poured from a bottle and used as food (e.g., salad oils).⁶⁸ A pediatric RCT included a canola diet component,⁵² and a non-RCT provided children with a poorly defined hypoallergenic diet.⁷⁰ Each report failed to specify exact serving sizes and the amount of omega-3 fatty acids derived from these dietary elements, although one non-RCT report claimed, without data, that their diets were matched for energy intake.⁷⁰ The issue of uncontrolled servings/dosing is problematic for the following reason.

Methods of delivering exposures in an uncontrolled fashion preclude knowing exactly what the “exposure” is, and cannot help but lead to variation in individual participants' daily intake across a study. This, in turn, translates into daily changes in a study group's daily consumption, or “exposure,” across a study, thereby complicating the interpretation of study results. The exposure is constantly changing within- and between-participants, which is quite different from the situation where a pediatric RCT weight-adjusts its on-study EPA/DHA doses.⁶⁴

In the few instances where consumption data were reported in studies employing uncontrolled servings, they merely illumined that, on average, participants either did not consume the mandated amount, or, in the case of a study employing an instruction pertaining to a range of possible consumption, the maximum allowable amount.^{68, 71} Nonetheless, a precise and constant definition of the treatment is required to meaningfully interpret study results, and the present treatment studies employing uncontrolled dosing strategies did not achieve this ideal.

Where controlled studies are concerned, uncontrolled servings created other problems, given that control participants also had their exposure delivered in the same, uncontrolled manner. This made it highly unlikely that participants in different study groups received the same amount of oil- or food-as-calories,⁶⁶ or that the difference in the amount of omega-3 fatty acid content consumed in the two study arms —reflecting a planned disparity (e.g., 5.4g/day from fish oil vs virtually no g/day from other oils)— was kept constant. Thus, unlike studies of controlled dosing (e.g., identical capsules containing fish oil or olive oil), and notwithstanding compliance, uncontrolled serving/dosing studies fail to provide a precise and constant definition of the exposure as oil/food or omega-3 fatty acid content. The ability to unequivocally interpret study results is thus hindered, while raising serious concerns about the internal validity of the three treatment studies conducted in Japan with perilla seed supplementation and which never failed to find a significant clinical effect.^{59, 66, 71}

The third confounding factor relates to the impact on results of the on-study use of corticosteroids. It is important to know exactly how many corticosteroid users were included in a given study, at what doses, and whether or not these doses were changed (and how) due to improved or failing asthma control. Yet, knowing these patterns of use will not necessarily make it less difficult to meaningfully interpret study results. The reason is that corticosteroid users may have had their doses altered across a study, or, in controlled investigations the distributions of users and of doses across study groups may have been unequal. In these circumstances, the ability of corticosteroids to improve respiratory outcomes particularly over the longterm can mask the benefits associated with omega-3 fatty acid use. For example, in sensing a lessening of their asthma control, participants receiving a placebo in an RCT might have their corticosteroid dose increased. This could improve respiratory functioning to a level equal to that produced by the omega-3 fatty acids intervention given to the other study group. A lack of cross-arm equivalence, either produced because of a change in one study arm, or because more corticosteroid users were enrolled in the control arm to begin with,⁶⁷ could eliminate any between-group differences in outcome that would denote a significant clinical effect in favor of the omega-3 fatty acids exposure. Also, within a single group of participants, the effects of increased doses of corticosteroids can bring about improved respiratory functioning thought to be attributable to the omega-3 fatty acids. Failing to know the exact role played by corticosteroids in a study owing to inadequate study design, reporting, or both, makes it impossible to rule out their possible impact on study results.⁵²

In the relevant studies of all design types, there was a scarcity of information regarding users of corticosteroids and their doses, including to which arms users had been allocated.^{57, 62, 64} Even less information was provided concerning whether doses of these drugs were kept constant or how they may have changed across the study for participants; or, whether the number of users and their doses were equivalent across study groups in controlled investigations. Yet, one RCT did ask participants to maintain a constant use of inhaled corticosteroids, although compliance data were not reported.⁶⁷ Another trial reporting no clinical effects claimed that similar oral corticosteroid doses had been observed for the two study arms.⁵⁴ Finally, having all participants not take corticosteroids might provide one of the clearest tests of the benefits of omega-3 fatty acids,⁶⁹ especially since Emelyanov et al.'s RCT of corticosteroid-naïve adults also excluded current or ex-smokers, provided controlled dosing via capsules, controlled for the intake of calories across study arms, and randomized one of the largest samples (n = 46) identified by the present review.⁶⁹

Thus, the goal with respect to these three factors is to assess and control their confounding influences. Otherwise, it may be impossible to unequivocally attribute significant or null clinical effects to the impact of the omega-3 fatty acids. Given how poorly the present collection of studies fared in achieving this goal, it is difficult to place much trust in the internal validity of many of the treatment studies in spite of their good Jadad-defined quality. One RCT selected only nonsmokers, yet failed to report whether they had also obtained information allowing them to rule out participants' smoking history.⁶⁷ Another factor worth noting for its possible influence as a confounder, yet for which there was a similar lack of appreciation in the present studies, is background diet. Very few investigators asked participants to maintain a constant background diet to assure that changes thereto would not bring additional variation to the task of explaining results.^{57, 70}

On the other hand, the primary prevention studies were far more likely to recognize the need to control for at least two of these confounders (i.e., factors influencing asthma control, and specifying exposures)⁴⁸ although these investigations were not conducted without limitations of their own. Most of the observational studies adopted a cross-sectional perspective whereby the timeframes associated with the assessment of the frequency of dietary fish intake were likely too short (e.g., the last month, current intake)^{48, 76} to reliably shed light on the possible influence of lifetime dietary intake patterns on the risk of developing asthma.⁷⁸ Moreover, in that interview questionnaires likely produce less misclassification of food-related information than is found in self-administered surveys, results using the former strategy might have been quite different.⁴⁸ Unfortunately, Huang et al.'s interview data regarding PUFA use were not broken down by type of PUFA.⁴⁸

The primary prevention studies also investigated the possible impact of different definitions of fish. For example, some identified the types of (e.g., oily) fish rich in omega-3 fatty acids,⁷⁷ whereas at least one of the studies including adolescents did not make any distinctions regarding the type of fish.⁷⁶ This might account for the positive relationship between fish intake and asthma prevalence. Had only fish assumed to be rich in omega-3 fatty acids been investigated, the positive association might have disappeared. The other study with adolescents did differentiate by type of fish, yet their sample was small.⁴⁸ No studies directly assessed the possibility that the ways of preparing fish rich in omega-3 fatty acids (e.g., frying; salting) could influence results by altering their fatty acid content.⁴⁸ Finally, no study attempted a comprehensive analysis of the omega-3 fatty acid content of the fish participants had eaten, at best preferring to define “oily fish” as containing more than 2% fat, for example.^{76, 77} As with some of the treatment studies, the primary prevention RCT included, as part of its intervention, uncontrolled servings of margarine and oils.⁵¹

The Decision to Forego Meta-Analysis

A number of criteria required satisfaction before meta-analysis could be considered. First, only RCTs were eligible, because of their greater potential to control for certain biases in meaningfully elucidating questions about the efficacy of any intervention/exposure. Second, it was established that, given the differences in clinical picture associated with age, data from pediatric and adult studies should not be combined in qualitative or quantitative synthesis. At the same time there had to be at least two studies with data capturing the same outcome construct (e.g., asthma severity). Yet, it is likely the case that having data contributed by more than two studies is best, especially if the studies contain small samples. More reliable estimates of variation may be derived.

Regarding Questions 1 (improving respiratory outcomes), 2 (impact of effect modifiers), and 3 (influencing mediators of inflammation), there were only seven adult and two pediatric treatment trials; and, meta-analysis with data from children was impossible given that the RCTs neither employed the same clinical or intermediate (i.e., mediators of inflammation) outcomes. One RCT had to be excluded from consideration for quantitative synthesis since it did not provide demographic information sufficient to determine even the age of participants.⁶⁵

The varying definitions of symptom scores observed in three RCTs ruled out any possibility of meta-analysis;^{57, 58, 67} and, while there were four separate occasions when at least four RCTs employed the same clinical outcome (FEV₁, AM PEF, PM PEF, bronchodilator use), several important observations translated into a decision to forego meta-analysis.

First, to meaningfully compare, then combine, RCTs evaluating the possible impact of omega-3 fatty acids, the contrasts involving a treatment and comparator should be similar or the same. In five of seven RCTs, the contrast involved controlled dosing of fish oils (i.e., EPA/DHA) compared with similar olive oil capsules.^{57, 58, 67–69} One other RCT compared high- versus low-dose EPA ethyl ester, while a seventh compared uncontrolled servings of perilla seed oil and corn oil.⁶⁶ The interpretability of a pooled estimate combining data from one of the latter two trials with any of the five studies comparing fish oil and an olive oil control would be limited. The contrasts are too dissimilar. This restricts the pool of studies.

Second, as discussed previously, there was too much missing, limited or contradictory information concerning both the adult study populations (e.g., asthma diagnosis details, including severity, duration, and concomitants/triggers) and study interventions (e.g., omega-3 fatty acid content).⁶⁶ Moreover, in studies of older adults, and for whom descriptions of current or past smoking were typically unavailable, there is little confidence that trialists reliably excluded adults with COPD.^{54, 58, 66} The situation that these examples illustrate does not permit an unambiguous appreciation of the populations and interventions to whom any pooled results could be generalized. Combining studies whose specific potential for generalization is uniformly low, or that vary on this basis, would yield a single estimate without a well-defined target. In the present collection of studies, so many population and intervention data were missing, making it difficult to generalize any meta-analytic results.

Third, there were too many studies of adults with flawed or limited research designs and methodologies to be able to meaningfully determine the relative contributions of each of the wanted and unwanted influences (e.g., the above-noted confounders) on a pooled effect. Fourth, certain study populations were too different to consider combining data from them. For example, one adult trial randomized in-patients,⁶⁶ whereas all other samples included out-patients; and, one RCT studied allergic asthmatics during pollen season.⁶⁷ Also, there was likely strong heterogeneity in the background diets given the countries in which the studies were conducted. Overall, while there were sufficient numbers of RCT to consider, a decision was made to forego any meta-analysis.

Regarding the primary outcome, FEV₁, three very different intervention-comparator contrasts were represented. They included: uncontrolled perilla seed oil supplementation compared with uncontrolled corn oil supplementation;⁶⁶ a high as opposed to a low dose EPA ethyl ester;⁵⁴ and, EPA/DHA from green-lipped mussel versus olive oil.⁶⁹ This lack of comparability in contrasts, combined with other key differences (e.g., in-patients in a very structured environment who received the uncontrolled supplementation;⁶⁶ very different background diets) made it inappropriate to pool their data. The situations with respect to the other three outcomes are somewhat more complicated.

Five studies reported having collected AM PEF data, yet one did not report data.⁵⁸ As well, Okamoto et al. gave uncontrolled servings of oil to in-patients,⁶⁶ while Stenius-Aarniala et al. used uncontrolled servings of oil in their 3-phase crossover RCT.⁶⁸ This intervention-related problem complicates the inclusion of these two studies in any synthesis. Stenius-Aarniala also failed to report inclusion criteria and, whether corticosteroid use was balanced across study arms.⁶⁸ They also did not employ a washout, and thus control for a possible carryover effect; and, they never established blinding given there was no attempt to conceal the taste of the dietary oil supplementation. Furthermore, Stenius-Aarniala et al. acknowledged having randomized three smokers and 12 ex-smokers, yet never made it clear that they had ruled out COPD, or that these participants were equally distributed across the study arms. Finally, seven participants withdrew due to problems with the taste of the oil, although there was no indication of the study arm from which they withdrew. This RCT was extremely flawed, and received the lowest Jadad total quality score of all trials. Their total quality score of 2 indicates low internal validity.

Also with respect to AM PEF, Thien et al.'s study on the one hand provided insufficient data regarding their population (e.g., asthma severity; diagnostic method), while at the same time suggesting that they had randomized a very unstable population (i.e., allergic asthmatics studied during and outside pollen season) which was very different from any other study population in the present review.⁶⁷ They also did not establish participants' smoker status, appeared only to begin to collect “pretreatment” data at the same time treatment began, and reported data only for completers. Almost a third of participants left the study due to nausea or vomiting, in addition to various difficulties swallowing the 18 capsules per day. They nevertheless received a Jadad total quality score of 4 on the strength of solid blinding, and a clear description of participants lost to followup.

Arm et al. failed to report inclusion or exclusion criteria, the method by which the diagnosis was determined, participants' smoker status, and whether the use of inhaled corticosteroids was balanced across the study arms.⁵⁷ They also reported three dropouts due to problems with the size and number of capsules. Finally, notwithstanding their use of a unique source of EPA/DHA (i.e., green-lipped mussel), Emelyanov et al.'s RCT was strong methodologically.⁶⁹ All studies that obtained AM PEF data were conducted in different countries, indicating a wide divergence in background diets.

Thus, on the basis of the aforementioned limitations and problems, as well as their lack of comparability, meta-analysis of AM PEF data was not considered appropriate. All of these same studies also provided either PM PEF^{57, 58, 67–69} or bronchodilator use data,^{57, 58, 67, 69} thereby making inappropriate any quantitative pooling of these datasets. Three of these studies contributed data pertaining to the impact of omega-3 fatty acid supplementation on mediators of inflammation.^{54, 57, 66} On the basis of divergent intervention-comparator contrasts and the above-noted problems, meta-analysis was not conducted.

A recent Cochrane review exclusively of fish oils in asthma conducted a number of meta-analyses of respiratory outcomes.³⁵ As with the decision made in the present review, they did not pool studies with different intervention-comparator contrasts. On the other hand, they did enter the Dry and Vincent trial data⁶⁵ into meta-analysis in spite of a lack of population-related information that would have permitted this RCT's classification as an adult or pediatric investigation. Moreover, they pooled data from this trial with that from a pediatric RCT,⁵² and failed to find a significant effect for FEV₁. While they performed numerous analyses for PEF (undefined), including for the beginning and the end of studies, their analysis of change from baseline data revealed a nonsignificant effect. PEF data from one pediatric trial⁵² were pooled with those from three adult study reports.^{56, 57, 67} In doing so, Woods et al.³⁵ likely entered duplicate data since the two Arm et al. trial reports included a large number of the same participants.^{56, 57} Analysis of asthma symptom scores included data from scales assessing varying constructs, and revealed no effect. Data from two pediatric studies^{52, 64} and three adult study reports^{56, 57, 67} were pooled, including those from the Arm et al. publications.^{56, 57} Heterogeneously defined asthma medication data were then combined, again without finding a significant effect. One pediatric⁵² and the two Arm et al. datasets^{56, 57} were included with results from Thien et al.⁶⁷ Data pertaining to varying definitions of bronchial hyper-responsiveness likewise failed to reveal a significant benefit. Results from two studies randomizing children,^{52, 64} the Thien et al. trial,⁶⁷ and from the Arm et al. publications^{56, 57} were combined. When bronchial hyper-responsiveness results from children and adults were meta-analyzed separately, no benefits were found for fish oil supplementation. Nonsignificant effects were also reported for asthma symptom score results when pediatric and adult data were analyzed independently.

Whether or not the Cochrane review's results suggest what the present review might have found had meta-analysis been conducted for respiratory outcomes, it is clear that the Cochrane undertaking failed to critically assess included studies so as to seriously consider foregoing meta-analysis, appeared to enter duplicate data, pooled results from children and adults, and included data from a trial without any specification of basic information such as age. Even so, they reported no benefit associated with fish oil supplementation.

Meta-analysis of primary prevention data was not considered, given the existence of one RCT.⁵¹ Too few safety data were observed to consider their further synthesis. Overall, poor reporting practices, which led to an inability to know whether and how these or other confounders might have influenced individual treatment RCT results, together with the lack of comparability in many of the RCTs' parameters (e.g., intervention-comparator contrasts), led to the decision to forego meta-analysis. Any pooled estimates would have been derived within a context instilling as little confidence in the appropriateness of the extrapolations of results as in the validity of the results themselves.

Clinical Implications

For those participants entered into treatment (or primary prevention) studies there is no notable safety profile associated with the consumption of omega-3 fatty acid content. As well, there is no consistent evidence, or any evidence from well-designed and sufficiently powered studies to recommend their use or avoidance to treat asthma in populations in North America, or in those countries in which their efficacy was tested. There is thus no way to conclude anything definitive about their therapeutic potential. Moreover, to which factors (e.g., type, source or dose of omega-3 fatty acids) their value as a therapy can be attributed cannot be established based on the present evidence. While some studies did investigate the impact of high doses,⁵⁴ it is clear that more studies of this sort, albeit with more than 12 participants, must be conducted. Various sources (e.g., marine, seed) and types of omega-3 fatty acid content (e.g., EPA, EPA/DHA, ALA) were also investigated, but they too require proper testing in trials that are larger and better-controlled. At present, from treatment studies, perhaps the only consistent observation regarding the short- or longterm use of omega-3 fatty acids with the types and doses herein evaluated is that it is unlikely to do notable harm.

While the results obtained by Emelyanov et al. are interesting because they gave a relatively large adult sample a low dose of a unique marine source (i.e., 200mg/d of extract of green-lipped mussel, with 400 mg/d olive oil vs 600mg/d olive oil), while also controlling several confounders (i.e., corticosteroid-naïve; no current or ex-smokers; controlled dosing; equal intake of calories across study arms), more research is required to establish that their four (of seven) significant clinical benefits did not occur because of some undetermined factors (e.g., blinding broken). Likewise, more work is required to understand how three studies of invariably limited intervention length (≤4 weeks) failed to produce a nonsignificant clinical effect when uncontrolled doses of perilla seed oil, and thus undefined amounts of ALA, were given to adults.^{59, 66, 71} Further testing might show that the observed clinical effects diminish over longer periods of time, the consumed doses of ALA were extremely high, or, in the case of two of these studies, the results were obtained because in-patients in controlled environments had been enrolled.^{59, 66} To better understand the possible utility of this type/source of omega-3 fatty acids, more research is needed.

One other factor that needs to be accounted for with respect to the perilla seed oil supplementation is that all three studies were conducted in Japan, where the omega-6/omega-3 fatty acid ratio in the typical diet (4:1) is much lower than in the United States (10–30:1), for example.^81–83 The lower ratio in Japan might make it more likely that this country's population can be affected by additional omega-3 fatty acid supplementation. With “less competition” from AA for the same metabolic pathways, EPA and DHA may be better able to affect the inflammation-based process by which the symptoms of asthma are produced. Also, it may explain why, in spite of poor air quality and higher rates of smoking,⁸⁴ for example, asthma is less prevalent in Japan (0.7%) than it is worldwide (5%).⁸⁵ As an aside, a crude assessment of the numbers of significant result relating to respiratory outcomes showed that clinical benefits accruing to omega-3 fatty acid supplementation were more likely in RCTs (6/6 in two trials^{64, 66} vs 6/38 in the remaining eight trials) and noncomparative case series investigated in Japan (13/15 in three noncomparative case series^{59, 62, 71} vs 15/35 in the other 6 studies) than in all other countries combined.

At the same time, it may be the case that the continuing high consumption of omega-6 fatty acids in countries such as the United States can offset the possible asthma-related benefits from omega-3 fatty acid supplementation. Some have even speculated that a highly imbalanced omega-6/omega-3 intake ratio in favor of the omega-6 fatty acids predisposes individuals to asthma.³⁰ Correct or not, it is likely essential that background diet be accounted for in the interpretation of individual study results, as well as any between-study differences.²⁴ In trying to explain the impact of PUFA consumption in asthma, it is also likely important to take into consideration the possibility that populations may vary in terms of the genetic factors influencing the expression of the asthma phenotype. It may also be best in planning treatment studies, as was seen in a few included studies^{70, 73} and the primary prevention RCT,⁵¹ to try to modify these PUFAs simultaneously.

Broughton et al. have suggested that it is the ratio of omega-6 to omega-3 fatty acid consumption that inhibits or attenuates inflammatory activity, leading to reliable respiratory benefits.⁷³ They have noted that modifying this ratio is critical for altering eicosanoid biosynthesis in rats. One implication is that, while providing omega-3 fatty acid supplementation alone necessarily alters the intake ratio involving omega-3 and omega-6 fatty acids, it is the active, marked reduction in omega-6 fatty acid ingestion that should likely accompany omega-3 fatty acid supplementation. With so few studies in the present evidence collection measuring background diet, however, study participants' intake of all PUFAs largely remains unknown.

Determining whether, and how, the clinical benefits of omega-3 fatty acid supplementation are made possible by their influence on mediators of inflammation was not an objective of the present systematic review. Nonetheless, an informal assessment is likely appropriate. There were few studies that observed significant and consistent influences of omega-3 fatty acids on mediators of inflammation (i.e., any leukotriene series), as well as too few studies that found significant and consistent clinical benefits while also having collected data regarding their impact on mediators of inflammation. Possible reasons for both observations include poorly designed studies, varying populations, and, small sample sizes. Before these observations are scrutinized further, however, an even more basic question may be whether there is evidence from the present collection of treatment studies that the omega-3 fatty acid exposures were incorporated into the biosystems of participants receiving them. EPA content did typically increase significantly although a concomitant, significant decrease in the AA content of tissue/plasma was not reliably observed.^{57, 67, 68, 74} But, this picture did not correlate well with significant clinical effects. Three of four nonsignificant clinical effects for AM PEF were associated with increased fatty acid content in tissue/plasma,^{57, 67, 68, 74} and, an RCT reported increased fatty acid content yet nonsignificant clinical effects for FEV₁ as well as five other respiratory outcomes.⁵⁴

Five studies were identified which employed one or more of the four respiratory outcomes (FEV₁, AM PEF, PM PEF, bronchodilator use) meeting the criteria to address Question 2 (i.e., at least two studies, with at least one demonstrating a significant clinical effect), while also reporting data with respect to the impact of omega-3 fatty acids on mediators of inflammation. Three were RCTs^{54, 57, 66}) and two were noncomparative case series.^{59, 71} The only pattern highlighting a mediator of inflammation that was exclusively associated with a significant clinical effect involved the suppression of LTC₄. That is, when LTC₄ generation was significantly suppressed, two RCTs and one noncomparative case series reported a significant clinical effect (i.e., increase) for AM PEF.^{59, 66, 71} Interestingly enough, all three studies had used uncontrolled perilla seed supplementation. Also, when perilla seed supplementation led to a significant suppression of both LTB₄ and LTC₄ produced by leukocytes, two of these same studies reported a significant increase in AM PEF,^{59, 66} and there was a significant effect for FEV₁ in one of these trials.⁶⁶ Yet, two RCTs observed no benefit from fish oil supplementation despite a substantial attenuation of neutrophil chemotactic responses to LTB₄, fMLP, and C5a.^{54, 57} Unfortunately, given the limitations of these studies (e.g., small samples), all that these observations can suggest are possible future directions for research.

With the present studies, it is thus impossible to adequately address the issue of the observed lack of “translation” into significant clinical effects of some of the observed changes in fatty acid content in tissue/plasma or the observed influences on mediators of inflammation. Future research with asthmatic participants may discover that the present servings/doses of omega-3 fatty acids had been too small, mediators associated with pathways other than the lipoxygenase and cyclooxygenase ones are actually more important to the pathogenesis or treatment of asthma,⁷⁸ or that without also intentionally altering the intake of omega-6 fatty acids, positive impacts on both the mediators of inflammation and respiratory outcomes may be unlikely. Of note, the small numbers of study identified by this review made it impossible to adequately assess the relationship between the less “potent” LTB₅, and clinical effects.⁸⁶ One small RCT (n = 12) in which a significant increase in LTB₅ was observed, produced a nonsignificant effect for FEV₁.⁵⁴ Finally, the lipid mediators of inflammation were primarily evaluated in the present evidence collection; and, nonsignificant results regarding TNF-a were obtained from a pediatric study.⁵²

One of this report's peer reviewers highlighted a very recently published study that, because of its late arrival, could not be systematically reviewed in the present review. Nevertheless, it is summarized here because it identified another exposure that might prove interesting in future studies of the impact of omega-3 fatty acids on leukotriene biosynthesis. In a three arm trial, 43 adults with mild to moderate, atopic asthma were randomized to receive, for 4 weeks, either 10 g of an emulsion containing 0.75 GLA and 0.5 g EPA, 15 g of this emulsion (1.13 g GLA and 0.75 g EPA), or, an olive oil placebo.⁸⁷ Results indicated a significant increase in plasma levels of EPA, DHA, dihommogamma-linolenic acid, and GLA; and, relative to placebo, stimulated whole blood LTB₄ biosynthesis decreased significantly in both active study arms. What is interesting is that GLA and EPA derive from different “parent” PUFAs.

It may also be too soon to conclude with respect to asthma, that omega-3 fatty acid supplementation is a better primary prevention than a therapeutic. While the results exclusively with children suggest a protective role, the studies involving adolescents do not. However, the failure to distinguish consumption data by type of fish in one study,⁷⁶ and the very small sample in the other,⁴⁸ might account for the positive associations of fish intake and asthma prevalence for adolescents. The adult study⁷⁵ found no relationship, but this study also did not distinguish by fish type in the same way that one pediatric study had, for example.⁷⁷ Interestingly enough, the other study involving children likewise did not draw such distinctions, yet it still reported a significant negative correlation between fish intake and asthma prevalence.⁷⁸ The reason for this result is unknown. Unfortunately, none of the studies reported having guesstimated the omega-3 fatty acid content of their subjects' exposures. What remains to be seen is whether or not the speculation that the capacity to alter risk may be inversely related to age is supported by future research. What appears certain is that a strong assessment of the possible protective role of omega-3 fatty acids in early childhood is currently underway.⁵¹

Two additional, recently published studies were pointed out by the above-noted peer reviewer. While they also arrived too late to be systematically reviewed, their key observations are likely worth mentioning. Nafstad et al. prospectively evaluated a cohort of Norwegian children and found that an inverse relationship between the early dietary intake of (any) fish (i.e., in the first 12 months of life) and the risk of asthma at age 4 years (n = 2,531) was only statistically significant in the bivariate analysis.⁸⁸ Adjusting for many factors such as parental atopy and respiratory tract infections, multivariate analysis did not reveal a statistically significant association. This appears to confirm the collective observation from the aforementioned primary prevention studies with children. Finally, Woods et al.'s adjusted and unadjusted analyses demonstrated a lack of association in young Australian adults (aged 20–44 years; n = 1,601) between fish intake (types undefined) and asthma risk.⁸⁹ This seems to confirm Troisi et al.'s finding from the Nurses study.⁷⁵ Since neither of these studies was systematically reviewed, their key findings must be taken with caution.

It may turn out that the most profound protective effect requires the incorporation of omega-3 fatty acids in one's weekly diet beginning early in life. It may also be found that the propensity for eating fish starting early in life is naturally associated with a lesser intake of foods rich in omega-6 fatty acids. Yet, as children age, and with the increasing influence of a less balanced intake of omega-3 and omega-6 fatty acids, the possibility for protection against asthma diminishes. Eventually, it may be observed that different ratios, or balances, of omega-6/omega-3 intake increase and decrease the risk of asthma.⁸⁴ The possible protective effect of non-marine sources (e.g., ALA) may be a question worth exploring as well.

Other studies have looked at the possible protective effects of dietary fish intake, yet each was excluded from the review because, instead of evaluating its relationship to asthma, it assessed possible associations with respiratory symptoms (wheeze, bronchitis)⁹⁰ or FEV₁ in adults over 29 years of age,⁹¹ respiratory symptoms (e.g., wheeze) in children 7 to 11 years of age⁹² or adults 20 to 44 years of age,⁹³ and, bronchial hyper-responsiveness in children and adults.⁹⁴ A final study was excluded because it did not specifically evaluate the possible impact of omega-3 fatty acids in their assessment of the relationship of dietary PUFA intake and asthma risk.⁹⁵

Research Implications and Possibilities

The research implications of the present findings are likely singular in pointing out the need for more research to address the questions of treatment and prevention. It is unlikely that attempts to try and explain the inconsistent results concerning treatment will ever yield a definitive answer to the question of efficacy. Moreover, such an attempt may be irrelevant if one of the key concerns is finding out whether omega-3 fatty acid supplementation can serve as an efficacious therapeutic for North American children and adults. Almost none of the included studies involved such populations with asthma. Asthma is an important health care problem, and it might be wise to plan some next research steps.

That there are no studies of secondary prevention may say as much about the difficulties inherent in planning a longterm study to evaluate the impact of omega-3 fatty acid supplementation on the progression of asthma as in conducting it. Such an undertaking presupposes knowing the exact nature and timing of the milestones (e.g., fundamental changes in respiratory functioning) marking the “natural progression” of asthma, and whose trajectory might be altered by omega-3 fatty acid supplementation (with or without modification of omega-6 fatty acid intake). At this time, what is known about the natural progression of asthma needs to be fully appreciated to permit designing such a study. On the other hand, the best test to date with respect to primary prevention is ongoing,⁵¹ and its results may be pivotal in identifying the dietary/lifestyle changes required to prevent the development of asthma in children at risk. Early intervention may turn out to be the most cost-effective strategy.

With respect to the subject of treatment, it may be useful to design a large, well-powered, multi-site RCT in North America comparing three study arms, and with stratification. The study would follow the CONSORT guidelines for reporting so that study quality could be adequately appraised, and the results effectively compared with those highlighted by other studies. Equal numbers of adult males and females would be randomized to receive controlled dosing (i.e., capsules) of fish oil (i.e., EPA/DHA), perilla seed oil (i.e., ALA), or controls (likely olive oil). The goal would be to evaluate the absolute efficacy of each of the omega-3 fatty acid interventions (vs placebo), as well as their comparative efficacy.

Appropriate methods to randomize participants as well as conceal these allocations would be employed. Double-blinding may require strengthening by altering the taste of the oils (e.g., peppermint flavoring).⁹⁶ Capsules would be identical. The investigators would need to be mindful of another key issue. The identity of the exposure, including the exact types and amounts of omega-3 fatty acids, must be clearly known. Different fish oils, or even the “same” fish oil from a single manufacturer across or within batches, may contain different ratios of EPA and DHA; and, if these fish oils vary in terms of their ratios of EPA to DHA, this is a possible confounder when the dose of omega-3 fatty acids is calculated as “EPA plus DHA.” This calls for an in-depth assessment of the composition and purity of each exposure. As well, the presence and nature of other active agents perhaps added to the oils would need to be established.

Stratification would be by dose (high vs low), with the number and contents of capsules taken by all participants used to control the amount of oil/calorie intake. Control oil would be added to active interventions, as needed, to produce the low dose. Pilot testing could establish reasonable definitions of high and low doses especially for the perilla seed oil intervention. Yet, whether the intervention contains perilla seed oil, or flaxseed oil, as others might recommend instead as a source of ALA, there remains an as yet unanswered question posed from the point of view of biology: given its status as a “parent” omega-3 fatty acid, and relative to EPA and DHA, could increased ALA intake possibly affect asthma?

Half of the participants randomized to each arm would also have their background diet adjusted based on an assessment of their lifetime (assessed by decade) intake of omega-6 fatty acids, including patterns of preparing all foods. The goal would be to establish a more balanced omega-6/omega-3 intake ratio. Monthly assessments of dietary intake would be conducted. The remaining half of participants would have their lifetime background diet assessed prior to the study and followed on a monthly basis. A request would be made of this second group of participants to maintain their prestudy diet across the study. A panel of experts could determine the appropriate definition of the modified omega-6/omega-3 intake ratio as well as the length of the intervention period needed to observe a meaningful clinical effect. Compliance for all participants would be monitored closely via monthly contacts. Stratification by dose and diet would allow secondary questions to be investigated.

Unfortunately, there is no shortage of candidate participants for inclusion in such a study. To qualify, each would have to receive a diagnosis of asthma following established professional criteria. Data would be obtained concerning concomitants/triggers with the potential to influence asthma control (e.g., allergies), as well as the duration and severity of asthma determined via pre-established professional criteria. The prestudy severity of each of the concomitants/triggers would be ascertained. A judgment regarding how well the prestudy asthma is being controlled by medication could be derived. Data concerning all prestudy asthma medications, and doses, would be collected and participants would be asked to maintain their prestudy regimen while on-study. They would also be asked to notify the appropriate study liaison should their on-study asthma status change or their medication require modification. A predetermined protocol would guide all decisions. A clear indication of possible differences among study arms for asthma- and asthma medication-related variables would be established at baseline, although controlling experimentally for either of these variables might make the research design too complex.

It might also be easier to exclude adults who are current and ex-smokers, thereby excluding adults with possible COPD. Otherwise, this would constitute another variable requiring control. Excluded would be individuals exhibiting bleeding or clotting disorders. The inclusion of equal numbers of males and females is suggested by the possibility that hormonally mediated genetic processes may make the male lung more susceptible to exposures, result in lung immaturity in utero, and yield a different picture of asthma than is observed in females.⁹⁷ Any parallel study of children would require distinguishing asthma and disorders of wheezing as well as weight-adjusting treatment doses, for example.

While multiple respiratory outcomes could be used, it is important to employ an established standard to assess pulmonary function. FEV₁ is a likely candidate for primary outcome yet other objective measures (e.g., PEF; bronchodilator use; health care utilization), and even one subjective measure (e.g., functional status), might be appropriate. Changes produced by supplementation would turn out to be respiratory outcome/function-dependent. Modifications to medication use would also be measured, and this might indicate either worsening or improving asthma. A panel of experts could determine the most pertinent ways to assess fatty acid compositions in tissue/plasma, as well as those mediators of inflammation that may be impacted by supplementation. In the active treatment arms, identifying those who do and those who do not respond with the significantly increased suppression of LTC₄ by leukocytes in response to omega-3 fatty acid supplementation may help predict those for whom the treatment does and does not produce some clinical benefit.⁷¹

The details of this or any other proposed study require consultation and collaboration. Moreover, if there is a belief that there may be some asthma-related benefit associated with taking omega-3 fatty acid supplementation, then some might argue that only by randomizing North Americans will results be observed that exhibit the degree of applicability required to elucidate the treatment of asthma in this population. Others might suggest that research also needs to be done to clearly ascertain whether or not North Americans and other populations vary sufficiently on any bases, including or beyond their patterns of omega-6/omega-3 fatty acid intake (e.g., cultural; racial-ethnic; environmental; genetic), to justify the above-noted divide when it comes to generalizing results. And, if significant empirical differences are not observed, it may become untenable to suggest that results from studies of those living outside North America cannot be generalized to North Americans. Still others might argue that, given the methodologic problems observed with respect to the present evidence base, it is more important at this point in time to conduct small, inexpensive trials to further clarify the mechanisms responsible for the putative beneficial effect of omega-3 fatty acids in asthma while also determining key data such as “the appropriate dose” required to yield reliable effects.

Limitations of the Review

The assessment of RCT quality was conducted using validated instruments; and, notwithstanding the uniformly “unclear” status of studies' handling of the concealment of allocation, the grades indicated good quality. However, in exclusively utilizing the four constructs provided by the Jadad and Schulz instruments, this review missed an opportunity to find that, using other constructs, the quality of the present collection of RCTs was actually low. If one were to measure quality in terms of each study's reporting of having controlled for confounders discussed previously (e.g., lack of cross-arm equivalence in asthma medication users and doses), most of the studies would have received a grade of “unclear” or “inadequate.” A failure to explicitly address the issues raised by these possible threats to internal validity means their confounding influences in the studies cannot be ruled out.

A similarly restricted definition of study quality likely characterized assessments of studies using designs other than an RCT. While the Downs and Black instrument from which the five items were selected, is a validated instrument, the fact that these typically small studies with many methodological limitations received a mean score likely indicating good quality is misleading. As with the evaluation of RCTs, a more comprehensive quality assessment of these other study designs (e.g., inadequate reporting of how, and if, asthma medication changed in a cohort across a study) likely would have yielded a very different picture. That this tool was insensitive to certain key quality issues required that this review's authors discuss these issues in considerable detail.

It is conceivable that identifying the country in which a study was conducted may have been all that was needed to assign an applicability rating, and may account for the observation that, on only one of 26 occasions did the independent assessors disagree. Nevertheless, the two applicability tools developed for the purposes of this review were never validated. More work is required to validate the assessment of this construct in systematic reviews.

Finally, while one publication reporting data from the NHANES II survey was captured and excluded because asthma per se had not been investigated,⁹⁰ another publication referring to results of this same, large study was missed when electronic and manual searches were conducted for this review.⁹⁷ It was discovered only after the present qualitative synthesis had been completed. Considered evidence pertaining to primary prevention, this study revealed that, after adjusting for age, gender, and race, fish intake (undefined fish types) in proportions per week for American children ages six months to eleven years of age neither predicted asthma nor wheeze.⁹⁷ This finding contradicts the picture of a protective effect seen for Australian and Japanese children.^{77, 78} Yet, it is consistent with the findings from the other American primary prevention study, which likewise reported a nonsignificant association between fish intake (undefined fish types) and asthma prevalence in adult nurses. The difference between the findings of this pediatric study and the others reported earlier^{77, 78} may be related to sampling methods or the possibility that the American diet contains enough omega-6 fatty acid content to offset the benefits of eating fish, even in children.

Conclusion

The present findings suggest that, with omega-3 fatty acid supplementation intended to influence asthma, there is little probability of harm beyond occasional mild discomfort. The most frequent troublesome events were produced by the delivery of the oils by large numbers and sizes of capsule. On the other hand, the lack of sufficiently consistent evidence, as well as a paucity of evidence from well-designed, well-conducted and adequately powered studies, suggests that no definitive conclusion can yet be drawn regarding the efficacy of omega-3 fatty acid supplementation as a treatment for asthma in children or adults. Likewise, nothing specific can be concluded regarding the role of specific sources, types or doses of omega-3 fatty acid content in producing significant clinical effects. One possible explanation for the inconsistent findings is the heterogeneity in definitions of settings, populations, interventions/exposures, and the types and doses of asthma medication. To afford generalizability to adult and pediatric populations of North American asthmatic, or to those at risk, some research may need to be conducted on this continent. The present review highlighted some of the methodological issues worth considering in treatment RCTs. An interesting hypothesis requiring investigation relates to the possible asthma-related benefits associated with actively decreasing levels of omega-6 fatty acid intake concurrent with increasing the intake of omega-3 fatty acids.

Having too few well-designed studies with which to adequately address this question means that nothing definitive can be said about the influence of omega-3 fatty acids on those mediators of inflammation thought to be implicated in the pathogenesis of asthma, or about the actual role played by these mediators in asthma. More research is required.

No studies were identified which investigated the potential of omega-3 fatty acids as secondary prevention. Primary prevention attempts were found, again without unanimity in their findings. While two studies of children outside North America noted a protective effect of dietary fish intake for asthma, one recently identified American study reported no benefit. Moreover, studies outside North America and primarily including adolescents found that dietary fish intake actually increased the risk of asthma. The only study involving adults found no relationship between these variables. However, these studies employed varying sampling methods and definitions of both the frequency of fish intake and, fish types. Likely the most promising attempt to use omega-3 fatty acids as primary prevention involves a large, ongoing RCT of expectant mothers whose children at risk for asthma are being followed for five years. To date, 18-month, interim analysis data are too unreliable given the difficulties in diagnosing asthma in children this young.

At this point in time, aside from an acceptable safety profile, it is impossible to definitively conclude anything with respect to the value of using omega-3 fatty acid supplementation in asthma for adults or children either in or beyond North America.

Search Strategy 1

1. exp Asthma/

2. Bronchial hyperreactivity/

3. asthma$.mp.

4. wheez$.mp.

5. respiratory sounds/

6. exp LEUKOTRIENES/

7. leukotrien$.mp.

8. (lung$ or pulmon$ or respirat$).mp. [mp=title, abstract, cas registry/ec number word, mesh subject heading]

9. exp INFLAMMATION/

10. exp Inflammation Mediators/

11. inflammat$.mp.

12. 8 and (or/9–11)

13. or/1–7,12

14. exp fatty acids, omega-3/

15. fatty acids, essential/

16. Dietary Fats, Unsaturated/

17. linolenic acids/

18. exp fish oils/

19. (n 3 fatty acid$ or omega 3).tw.

20. docosahexa?noic.tw,hw,rw.

21. eicosapenta?noic.tw,hw,rw.

22. alpha linolenic.tw,hw,rw.

23. (linolenate or cervonic or timnodonic).tw,hw,rw.

24. menhaden oil$.tw,hw,rw.

25. (mediterranean adj diet$).tw.

26. ((flax or flaxseed or flax seed or linseed or rape seed or rapeseed or canola or soy or soybean or walnut or mustard seed) adj2 oil$).tw.

27. (walnut$ or butternut$ or soybean$ or pumpkin seed$).tw.

28. (fish adj2 oil$).tw.

29. (cod liver oil$ or marine oil$ or marine fat$).tw.

30. (salmon or mackerel or herring or tuna or halibut or seal or seaweed or anchov$).tw.

31. (fish consumption or fish intake or (fish adj2 diet$)).tw.

32. diet$ fatty acid$.tw.

33. or/14–32

34. dietary fats/

35. (randomized controlled trial or clinical trial or controlled clinical trial or evaluation studies or multicenter study).pt.

36. random$.tw.

37. exp clinical trials/ or evaluation studies/

38. follow-up studies/ or prospective studies/

39. or/35–38

40. 34 and 39

41. (Ropufa or MaxEPA or Omacor or Efamed or ResQ or Epagis or Almarin or Coromega).tw.

42. (omega 3 or n 3).mp.

43. (polyunsaturated fat$ or pufa or dha or epa or long chain or longchain or lc$).mp.

44. 42 and 43

45. 33 or 40 or 41 or 44

46. 13 and 45

47. limit 46 to human

Search Strategy 2

#1 omega 3

#2 (“essential-fatty-acids” in SU) or (“linolenic-acid” in SU)

#3 (“docosahexaenoic-acid” in SU) or (“eicosapentaenoic-acid” in SU)

#4 explode “plant-oils” in SU

#5 explode “fish-oils” in SU

#6 “fish-consumption” in SU

#7 “polyenoic-fatty-acids” in SU

#8 “polyunsaturated-fats” in SU

#9 “dietary-fat” in SU

#10 (n 3 fatty acid* or omega 3) in ti,ab,id

#11 (docosahexanoic or docosahexaenoic) in ti,ab,id

#12 (eicosapentanoic or eicosapentaenoic) in ti,ab,id

#13 (alpha linolenic)in ti,ab,id

#14 (linolenate or cervonic or timnodonic) in ti,ab,id

#15 (mediterranean diet) in ti,ab,id

#16 ((flax or flaxseed or flax seed or linseed or rape seed or rapeseed or canola or soy or soybean or walnut or mustard seed or menhaden) and oil*) in ti,ab,id

#17 (walnut* or butternut* or soybean* or pumpkin seed*) in ti,ab,id

#18 (fish oil* or cod liver oil* or marine oil* or marine fat*) in ti,ab,id

#19 (salmon or mackerel or herring or tuna or halibut or seal or seaweed or anchov*) in ti,ab,id

#20 (fish consumption or fish intake) in ti,ab,id

#21 (diet* fatty acid*) in ti,ab,id

#22 (ropufa or maxepa or omacor or efamed or resq or epagis or almarin or coromega) in ti,ab,id

#23 ((omega 3 or n 3) and (polyunsaturated fat* or pufa or dha or epa or long chain or longchain or lc*)) in ti,ab,id

#24 “long-chain-fatty-acids” in SU

#25 (fish and diet) in ti,ab,id

#26 ((fish and diet) in ti,ab,id) or (“long-chain-fatty-acids” in SU) or (((omega 3 or n 3) and (polyunsaturated fat* or pufa or dha or epa or long chain or longchain or lc*)) in ti,ab,id) or ((ropufa or maxepa or omacor or efamed or resq or epagis or almarin or coromega) in ti,ab,id) or ((docosahexanoic or docosahexaenoic) in ti,ab,id) or ((n 3 fatty acid* or omega 3) in ti,ab,id) or (“dietary-fat” in SU) or (“polyunsaturated-fats” in SU) or (“polyenoic-fatty-acids” in SU) or (“fish-consumption” in SU) or (explode “fish-oils” in SU) or (explode “plant-oils” in SU) or ((“docosahexaenoic-acid” in SU) or (“eicosapentaenoic-acid” in SU)) or ((“essential-fatty-acids” in SU) or (“linolenic-acid” in SU)) or ((diet* fatty acid*) in ti,ab,id) or ((fish consumption or fish intake) in ti,ab,id) or ((salmon or mackerel or herring or tuna or halibut or seal or seaweed or anchov*) in ti,ab,id) or ((fish oil* or cod liver oil* or marine oil* or marine fat*) in ti,ab,id) or ((walnut* or butternut* or soybean* or pumpkin seed*) in ti,ab,id) or (((flax or flaxseed or flax seed or linseed or rape seed or rapeseed or canola or soy or soybean or walnut or mustard seed or menhaden) and oil*) in ti,ab,id) or ((mediterranean diet) in ti,ab,id) or ((linolenate or cervonic or timnodonic) in ti,ab,id) or ((alpha linolenic)in ti,ab,id) or ((eicosapentanoic or eicosapentaenoic) in ti,ab,id)

#27 Bronchial hyperreactiv*

#28 Asthma*

#29 Wheez*

#30 Leukotrien*238

#31 explode “leukotrienes-” in SU

#32 “respiratory-hypersensitivity” in SU

#33 explode “asthma-” in SU

#34 Lung* or pulmon* or respirat*

#35 explode “inflammation-” in SU

#36 Inflammat*

#37 #34 and (#35 or #36)

#38 (explode “leukotrienes-” in SU) or (Leukotrien*) or (Wheez*) or (Asthma*) or (#34 and (#35 or #36)) or (Bronchial hyperreactiv*) or (explode “asthma-” in SU) or (“respiratory-hypersensitivity” in SU)

#39 #26 and #38

#40 “man” in od

#41 #40 and #39

Letter to Industry Representatives from the Three EPCs Investigating the Health Benefits of Omega-3 Fatty Acids

May 2, 2003

Dear _________,

I am writing on behalf of the Evidence Based Practice Centers at RAND, New England Medical Center and the University of Ottawa. We are conducting a systematic review of the efficacy and toxicity of omega-3 fatty acids in the prevention and treatment of a number of different diseases/conditions. This review is being conducted under a contract from the Agency for Healthcare Research and Quality (AHRQ).

We are contacting you to see if there is any evidence, including unpublished evidence, that you want considered. Our focus is on clinical trials of omega-3 fatty acids in humans, so animal and chemical studies are not necessary.

The specific questions that all the EPCs will address are detailed in the attachment to this letter.

Please contact me with any information that you might have. I will be out of town next week and will respond to any questions when I get back. If you have any questions that you would like addressed before I return, please contact Donna Mead at the address above.

Best regards,

Catherine MacLean, M.D., Ph.D.

RAND1700 Main Street, M 23-C

Santa Monica, CA 90407-2138

Voice: 310 393-0411, x6364

Fax: 310-451-6930

maclean@rand.org

Relevance Assessment Form

Please respond to each of the first 7 questions. Comments typed into the box can identify duplicate reports, a key review whose references should be checked, anomalies, etc.

1. Inclusion criteria:

   • 1

     Does this study involve human participants?

     YES Can't Tell NO

   • 2

     Does this study employ (foods known to contain) omega-3 fatty acids (n-3) as an intervention/exposure?

     YES Can't Tell NO

   • 3

     Is the purpose of the intervention/exposure: a. when used by individuals with asthma, to improve respiratory outcomes, or, influence mediators of inflammation; or, b. to prevent asthma?

     YES Can't Tell NO

2. Exclusion criterion:

   • 4

     Is this a narrative or systematic review, opinion piece or editorial, letter, guideline or policy paper, etc. that, if it does so at all, only describes studies (and their results) reported elsewhere (i.e., it does not present evidence published for the first time)?

     YES Can't Tell NO

3. Context:

   • 5

     The study appears to investigate (select at least one option; click on all that apply):

     -n-3's potential impact on respiratory outcomes in individuals with asthma

     -n-3's potential impact on mediators of inflammation (e.g., leukotrienes) in individuals with asthma

     -n-3's potential to prevent asthma

     -children or adolescents as the target population

     -at least one control or comparator group

     -US population dietary intake data regarding n-3 or n-6 fatty acids

     -none of the above

   • 6

     The study appears to also or instead concern omega-3 fatty acids as an intervention/exposure for the following human health/disease domains (select at least one option; click on all that apply):

     -cardiovascular (human or non-human/in vitro focus)

     -gastrointestinal/renal

     -autoimmune

     -cancer

     -neurology

     -child/maternal health

     -immune-mediated

     -transplantation

     -mental health

     -none of the above

     -eye health

   • 7

     Is this study written up in a language other than English?

     YES NO

   • 8

     Comments box

Data Abstraction Form

Instructions: Please answer each question. Selecting response options means clicking on them. A text box requires you to provide specific information. When it is not reported (= NR), the question does not apply (= N/A), you cannot tell what/where it is (= CT), or you have no comment to make (= NC), type the relevant code in the text box. If the research report describes more than one study, answer in this eForm all the questions for the first reported study while at the same time letting the review manager know that another data abstraction form is required.

1. Initials of reviewer: COMMENTS BOX = BOX

2. Reference identification # (Refid#): BOX

3. Author, Year: BOX

4. Unique identifier # (type NR for now): BOX

5. Number of unique, review-relevant studies that this report describes: BOX

6. Publication status (select one):

   • Peer-reviewed journal publication

   • Journal publication

   • Conference abstract/poster

   • Book

   • Book chapter

   • HTA/technical report

   • Thesis

   • Unpublished document

   • Study sponsor's internal report

   • Internet document

   • Other

7. If you answered ‘Other’ to the preceding question, specify what you mean: BOX

8. If other review-relevant reports refer to this same study, provide their Refid #s: BOX

9. Country in which the study was conducted (select all that apply):

   • Australia

   • Canada

   • United States

   • Japan

   • United Kingdom (not Ireland)

   • France

   • Germany

   • Italy

   • Finland

   • Russia

   • Other

   • Not reported

10. If you answered ‘Other’ to the preceding question, specify what you mean: BOX

11. Number of sites: BOX

12. Funding source type (select all that apply):

    • Government

    • Industry

    • Private (non-industry)

    • Hospital

    • Other

    • Not reported

    • Can't tell

13. Specify the funding source(s): BOX

14. Study complexity (select one):

    • Multiple study arms, cohorts, or phases (i.e., crossover)

    • Single study arm/cohort

15. Study design (select one):

    • RCT: parallel design

    • RCT: cross-over design

    • RCT: factorial design

    • Controlled clinical trial (non-RCT)

    • Multiple cohorts

    • Single cohort

    • Case-control

    • Cross-sectional

    • Case series

    • Case study

    • Other

16. If you answered ‘Other’ to the preceding question, specify what you mean: BOX

17. Identify any notable details concerning (e.g., restricted randomization; blocking size), or problems with, the research design or its implementation (e.g., inappropriateness of: placebo/control(s), run-in and washout protocols/durations, etc.): BOX

18. Total # of individuals screened: BOX

19. # enrolled/randomized participants (hereafter, ‘study participants’): BOX

20. # study participants completing the study: BOX

21. Study participants' percentage of males: BOX

22. Comments, including notable differences between study arms/cohorts re ‘% male participants’: BOX

23. Mean age (SD/SE; range) of study participants: BOX

24. Comments, including notable differences between study arms/cohorts re age: BOX

25. Body weight of study participants (mean; range): BOX

26. Comments, including notable differences between study arms/cohorts re body weight: BOX

27. From which racial groups were study participants drawn (select all that apply)?

    • Black/African ancestry

    • Native American/Canadian

    • Inuit/Eskimo

    • Hispanic

    • Asian

    • Caucasian/European

    • Other

    • Not reported

    • Can't tell

28. Specify each racial group's percentage/proportion of full sample: BOX

29. Comments, including notable differences between study arms/cohorts re racial composition: BOX

30. Concurrent conditions (select all that apply)?

    • Diabetes

    • Lipid-related problems (e.g., hypercholesterolemia)

    • Allergic rhinitis

    • Atopy (e.g., atopic dermatitis)

    • Other allergies or sensitivities (e.g., aspirin/ASA)

    • Other

    • None reported

31. If you answered ‘Other’ to the preceding question, specify what you mean: BOX

32. Specify the type and severity (mean; SD/SE; range: with units) of each concurrent condition, as well as how it was defined and diagnosed: BOX

33. Specify the percentage/proportion of the whole sample re each type of concurrent condition: BOX

34. Comments, including notable differences between study arms/cohorts re concurrent conditions: BOX

35. Specify pre-study medications or treatments for each concurrent condition, with dose/frequency: BOX

36. Comments, including notable differences between study arms/cohorts re pre-study medications/treatments: BOX

37. Describe any pre-study use of omega-3 fatty acid supplements (may constitute a retrospective evaluation of dietary intake): BOX

38. Comments, including notable differences between study arms/cohorts re the pre-study use of omega-3 fatty acid supplements: BOX

39. How was the dietary intake of omega-3 fatty acid usage evaluated (select all that apply)?

    • Nutritionist-administered quantitative food-frequency survey(s)

    • Nutritionist-administered semi-quantitative food-frequency survey(s)

    • Self-administered quantitative food-frequency survey(s)

    • Self-administered semi-quantitative food-frequency survey(s)

    • Parent-administered quantitative food-frequency survey(s)

    • Parent-administered semi-quantitative food-frequency survey(s)

    • Direct measurement(s) of food intake

    • Survey(s), yet no details provided

    • Other

    • Can't tell

    • Not reported

40. If you answered ‘Other’ to the preceding question, specify what you mean: BOX

41. Describe any pre-study use of other dietary supplements, including dose/frequency: BOX

42. Comments, including notable differences between study arms/cohorts re the pre-study use of other dietary supplements: BOX

43. Types of pre-study diet attributed to study participants (select all that apply):

    • High fish diet

    • Fish-vegetarian diet

    • Low fish diet

    • Low fat diet

    • High fat diet

    • Mediterranean diet

    • Other

    • Can't tell

    • Not reported

44. If you answered ‘Other’ to the preceding question, specify what you mean: BOX

45. Specify percentage/proportion of participants on each diet: BOX

46. Comments, including notable differences between study arms/cohorts re the type of baseline diet: BOX

47. Describe the full sample's absolute fatty acid content of the baseline diet: BOX

48. Comments, including notable differences between study arms/cohorts re the absolute fatty acid content of the baseline diet: BOX

49. Describe the full sample's relative fatty acid content of the baseline diet: BOX

50. Comments, including notable differences between study arms/cohorts re the relative fatty acid content of the baseline diet: BOX

51. Describe the full sample's baseline blood lipid biomarker levels, with units: BOX

52. Comments, including notable differences between study arms/cohorts re these blood lipid biomarker levels: BOX

53. Describe the full sample's baseline (serum, tissue, or cell membrane) levels of fatty acids: BOX

54. Describe the full sample's baseline omega-6/omega-3 tissue ratios of fatty acid: BOX

55. Describe any notable differences between the study arms/cohorts re their baseline fatty acids (serum, tissue, or cell membrane) or omega-6/omega-3 tissue ratios of fatty acid: BOX

56. What percentage/proportion of participants were diagnosed at study entry with asthma? BOX

57. Why were <100% of participants diagnosed with asthma at study entry (e.g., primary prevention study): BOX

58. Was asthma clearly defined? YES CT NO NR

59. How was asthma defined? BOX

60. How was asthma diagnosed? BOX

61. Identify the percentage/proportion of participants per asthma sub-type: BOX

62. According to the report, what was the likely cause(s) of the asthma (e.g., exercise- or aspirin-induced): BOX

63. Comments, including notable differences between study arms/cohorts re the definition, subtypes, or causes of asthma: BOX

64. What is the certainty of the asthma diagnosis? (select one)

    • Not applicable

    • No information reported

    • Can't tell

    • Possible: compatible symptoms

    • Probable: MD clinical diagnosis

    • Definite: MD clinical diagnosis or compatible symptoms PLUS objective measure of airway reactivity (FEV₁ improves ≥ 15% post-bronchodilator; methacholine challenge test reveals PC20 <8mg/ml; diurnal variation in PEF or FEV₁ > 15%)

65. Comments, including notable differences between study arms/cohorts re the certainty of the asthma diagnosis: BOX

66. Was asthma severity clearly defined? YES CT NO NR

67. How was asthma severity defined (e.g., daily symptoms of cough, wheeze, breathlessness; frequency of rescue inhaler use; nocturnal awakenings; exercise limitation; emergency visits and hospitalizations; recent use of systemic [oral; IV] corticosteroids; FEV₁, peak flow; peak flow variability; measured nonspecific airway reactivity [e.g., histamine or methacholine challenge testing])? BOX

68. Describe the full sample's baseline (mean; SD/SE; range) level of asthma severity: BOX

69. Comments, including notable differences between study arms/cohorts re participants' baseline severity of asthma: BOX

70. According to the authors, how well-controlled by medication(s) were participants' asthma? (select one)

    • Well-controlled

    • Symptomatic

    • Not indicated

    • Can't tell

    • Not applicable

71. How did the authors define well-controlled vs. symptomatic? BOX

72. Comments, including notable differences between study arms/cohorts re baseline asthma control: BOX

73. Describe any notable differences at baseline between study arms/cohorts in terms of any indices of participants' asthma clinical status or pulmonary function: BOX

74. Participants' asthma duration (mean; SD/SE; range): BOX

75. Comments, including notable differences between study arms/cohorts re asthma duration: BOX

76. Describe atopy status of participants, including their family history: BOX

77. Comments, including notable differences between study arms/cohorts re history of atopy: BOX

78. Participants' family history of asthma: BOX

79. Comments, including notable differences between study arms/cohorts re family history of asthma: BOX

80. Specify presence of furry/feathered pets at home or work (e.g., % participants): BOX

81. Comments, including notable differences between study arms/cohorts re presence of furry/feathered pets at home or work: BOX

82. Specify socioeconomic status of participants: BOX

83. Comments, including notable differences between study arms/cohorts re socioeconomic status: BOX

84. Describe participants' family size and living conditions (e.g., crowdedness; cleanliness): BOX

85. Comments, including notable differences between study arms/cohorts re family size or living conditions: BOX

86. Identify location of participants' present residence (select one):

    • 100% Urban

    • 100% Rural

    • Some from each

87. Comments, including notable differences between study arms/cohorts re residence location: BOX

88. # study participants (working) in daycare (for infections which exacerbate asthma): BOX

89. Comments, including notable differences between study arms/cohorts re study participants (working) in daycare: BOX

90. Describe participants' history of respiratory infections (e.g., bronchiolitis): BOX

91. Comments, including notable differences between study arms/cohorts re participants' history of respiratory infections: BOX

92. Describe participants' history of exposure to environmental tobacco smoke: BOX

93. Comments, including notable differences between study arms/cohorts re participants' history of exposure to environmental tobacco smoke: BOX

94. Describe participants' smoking history and present smoker status: BOX

95. Comments, including notable differences between study arms/cohorts re participants' smoking history and present smoker status: BOX

96. Other asthma risk factors, and percentage/proportion of participants re each (specify): BOX

97. Comments, including notable differences between study arms/cohorts re these risk factors: BOX

98. Other factors adversely affecting asthma control, and percentage/proportion of participants re each (specify): BOX

99. Comments, including notable differences between study arms/cohorts re these factors influencing asthma control: BOX

100. Describe pre-study asthma medication(s) or treatments, including dose/frequency: BOX

101. Comments, including notable differences between study arms/cohorts re pre-study asthma medication(s) or treatments, including dose/frequency: BOX

102. Comments re the appropriateness of pre-study asthma medications or treatments, including dose/frequency: BOX

103. Comments, including notable differences between study arms/cohorts re the appropriateness of the pre-study asthma medication(s) or treatments, including dose/frequency: BOX

104. Describe participants' pre-study exacerbation, emergency visit, or hospitalization rates: BOX

105. Comments, including notable differences between study arms/cohorts re the pre-study exacerbation, emergency visit, or hospitalization rates: BOX

106. Describe participants' pre-study rescue inhaler use: BOX

107. Comments, including notable differences between study arms/cohorts re the pre-study rescue inhaler use: BOX

108. Comments about any other asthma-related covariates (e.g., baseline serum IgE levels), including any notable differences between the study arms/cohorts in terms of these: BOX

109. Season the study was initiated: BOX

110. Season the study was completed: BOX

111. List the study's inclusion criteria: BOX

112. List the study's exclusion criteria: BOX

113. Intention of study (select all that apply)

     • Treatment

     • Primary prevention

     • Secondary prevention

     • Impact on mediators of inflammation

     • Safety

     • Other

     • Unclear

114. If you answered ‘Other’ to the preceding question, specify what you mean: BOX

115. Type of study (select one):

     • Interventional

     • Observational

116. Data were analyzed according to which criterion (select one)?

     • Intention-to-treat (all randomized/enrolled)

     • Those receiving at least one dose/serving

     • Those completing the study (i.e., with follow-up data)

     • Can't tell

     • Other

117. If you answered ‘Other’ to the preceding question, specify what you mean: BOX

118. Study duration, including units (includes run-in period duration, washout duration, etc.): BOX

119. Omega-3 fatty acid product name(s) (select all that apply):

     • Almarin

     • Coromega

     • Eiconol

     • Efamed

     • Epagis

     • MaxEPA

     • Menhaden oil

     • ResQ

     • Omacor

     • Ropufa

     • Other

     • Not reported

     • Not applicable

120. If you answered ‘Other’ to the preceding question, specify what you mean: BOX

121. Name of manufacturer of the omega-3 fatty acid product(s): BOX

122. Reported information re the purity of the omega-3 fatty acid product: BOX

123. Reported information re the presence of other, potentially active agents in the omega-3 fatty acid product: BOX

124. Asthma medications allowed or mandated during the study, including dose and frequency: BOX

125. Comments, including notable differences between study arms/cohorts/phases re participants' asthma medications, including dose/frequency: BOX

126. Permitted or mandated medications or treatments for concurrent conditions during the study (specify type, dose/frequency, and for which concurrent condition): BOX

127. Comments, including notable differences between study arms/cohorts/phases re participants' permitted or required medications or treatments for concurrent conditions: BOX

128. Review-relevant outcomes assessed (e.g., efficacy; incidence; prevalence; mediators of inflammation): BOX

129. Timing of follow-up assessment(s) per outcome, relative to start of intervention (e.g., after 8 hours; at week 4): BOX

130. Total # of study arms or cohorts (note: in a crossover trial, each phase type is considered an exposure/intervention ‘arm:’ e.g., placebo vs. omega-3): BOX

131. # crossovers: BOX

132. Define only the study arms, cohorts, or phases of interest to the present review (exclude others, for which data will not be extracted): BOX

133. With a specific omega-3 fatty acid ‘active’ arm in mind, type ARM1 in the text box: BOX

134. Re this study arm, what is the sample size at study entry? BOX

135. Sample size of those completing the study: BOX

136. Intervention length (e.g., weeks, months): BOX

137. Arm type (active; placebo; control) BOX

138. Intervention/exposure type (e.g., marine diet; plant diet; marine-based supplement; plant-based supplement; nuts; nut oil): BOX

139. Define the intervention/exposure components (e.g., ‘omega-3 rich diet;’ anchovy serving; fish oil supplement; flaxseed/linseed; rapeseed/canola; walnut oil; softgel capsule): BOX

140. Specific source(s) of omega-3 fatty acids (i.e., type[s] of fish or plant): BOX

141. Per-dose/serving amount of omega-3 fatty acids (e.g., grams): BOX

142. Dose/serving frequency (e.g., once daily): BOX

143. Timing (e.g., with breakfast; any time): BOX

144. Route of administration: BOX

145. Total daily (mean; SD/SE; range) omega-3 fatty acid intake (e.g., grams): BOX

146. Type(s) of omega-3 fatty acid identified (i.e., DHA, EPA, DPA, ALA): BOX

147. Total daily amount/dose (mean; SD/SE; range) of each type of omega-3 fatty acid, or combination (e.g., EPA+DHA) (e.g., grams): BOX

148. Omega-3 fatty acid composition (%) of the exposure/intervention: BOX

149. Permitted or mandated amount/dose (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure) of omega-6 fatty acid intake (e.g., grams): BOX

150. Permitted or mandated omega-6/omega-3 ratio of the exposure/intervention: BOX

151. Other permitted or mandated type(s) of co-intervention (e.g., anti-oxidants such as Vitamin E; other dietary supplements; other foodstuff), including respective dose/servings (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure), with units: BOX

152. Total daily fat intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

153. Total daily caloric intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

154. Describe whether, and why, it was necessary or possible to change the standard asthma medications/treatments during the course of the study: BOX

155. If standard asthma medications/treatments were changed, then how? BOX

156. If there is a second omega-3 fatty acid ‘active’ arm or, if none, then a control arm (e.g., placebo), type ARM2 in the text box (however, click here if there are no more arms/cohorts; this links directly to question #252) BOX

157. Re this study arm, what is the sample size at study entry? BOX

158. Sample size of those completing the study: BOX

159. Intervention length (e.g., weeks, months): BOX

160. Arm type (active; placebo; control) BOX

161. Intervention/exposure type (e.g., marine diet; plant diet; marine-based supplement; plant-based supplement; nuts; nut oil): BOX

162. Define the intervention/exposure components (e.g., ‘omega-3 rich diet;’ anchovy serving; fish oil supplement; flaxseed/linseed; rapeseed/canola; walnut oil; softgel capsule): BOX

163. If control/placebo, describe it (e.g., ‘omega-3 poor diet’) or what was used (e.g., olive oil, safflower oil): BOX

164. Specific source(s) of omega-3 fatty acids (i.e., type[s] of fish or plant): BOX

165. Per-dose/serving amount of omega-3 fatty acids (e.g., grams): BOX

166. Dose/serving frequency (e.g., once daily): BOX

167. Timing (e.g., with breakfast; any time): BOX

168. Route of administration: BOX

169. Total daily (mean; SD/SE; range) omega-3 fatty acid intake (e.g., grams): BOX

170. Type(s) of omega-3 fatty acid identified (i.e., DHA, EPA, DPA, ALA): BOX

171. Total daily amount/dose (mean; SD/SE; range) of each type of omega-3 fatty acid, or combination (e.g., EPA+DHA) (e.g., grams): BOX

172. Omega-3 fatty acid composition (%) of the exposure/intervention: BOX

173. Permitted or mandated amount/dose (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure) of omega-6 fatty acid intake (e.g., grams): BOX

174. Permitted or mandated omega-6/omega-3 ratio of the exposure/intervention: BOX

175. Other permitted or mandated type(s) of co-intervention (e.g., anti-oxidants such as Vitamin E; other dietary supplements; other foodstuff), including respective dose/servings (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure), with units: BOX

176. Total daily fat intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

177. Total daily caloric intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

178. Describe whether, and why, it was necessary or possible to change the standard asthma medications/treatments during the course of the study: BOX

179. If standard asthma medications/treatments were changed, then how? BOX

180. If there is an additional omega-3 fatty acid ‘active’ arm or, if none, then a control arm (e.g., placebo), type ARM3 in the text box (however, click here if there are no more arms/cohorts; this links directly to question #252) BOX

181. Re this study arm, what is the sample size at study entry? BOX

182. Sample size of those completing the study: BOX

183. Intervention length (e.g., weeks, months): BOX

184. Arm type (active; placebo; control) BOX

185. Intervention/exposure type (e.g., marine diet; plant diet; marine-based supplement; plant-based supplement; nuts; nut oil): BOX

186. Define the intervention/exposure components (e.g., ‘omega-3 rich diet;’ anchovy serving; fish oil supplement; flaxseed/linseed; rapeseed/canola; walnut oil; softgel capsule): BOX

187. If control/placebo, describe it (e.g., ‘omega-3 poor diet’) or what was used (e.g., olive oil, safflower oil): BOX

188. Specific source(s) of omega-3 fatty acids (i.e., type[s] of fish or plant): BOX

189. Per-dose/serving amount of omega-3 fatty acids (e.g., grams): BOX

190. Dose/serving frequency (e.g., once daily): BOX

191. Timing (e.g., with breakfast; any time): BOX

192. Route of administration: BOX

193. Total daily (mean; SD/SE; range) omega-3 fatty acid intake (e.g., grams): BOX

194. Type(s) of omega-3 fatty acid identified (i.e., DHA, EPA, DPA, ALA): BOX

195. Total daily amount/dose (mean; SD/SE; range) of each type of omega-3 fatty acid, or combination (e.g., EPA+DHA) (e.g., grams): BOX

196. Omega-3 fatty acid composition (%) of the exposure/intervention: BOX

197. Permitted or mandated amount/dose (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure) of omega-6 fatty acid intake (e.g., grams): BOX

198. Permitted or mandated omega-6/omega-3 ratio of the exposure/intervention: BOX

199. Other permitted or mandated type(s) of co-intervention (e.g., anti-oxidants such as Vitamin E; other dietary supplements; other foodstuff), including respective dose/servings (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure), with units: BOX

200. Total daily fat intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

201. Total daily caloric intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

202. Describe whether, and why, it was necessary or possible to change the standard asthma medications/treatments during the course of the study: BOX

203. If standard asthma medications/treatments were changed, than how? BOX

204. If there is an additional omega-3 fatty acid ‘active’ arm or, if none, then a control arm (e.g., placebo), type ARM4 in the text box (however, click here if there are no more arms/cohorts; this links directly to question #252) BOX

205. Re this study arm, what is the sample size at study entry? BOX

206. Sample size of those completing the study: BOX

207. Intervention length (e.g., weeks, months): BOX

208. Arm type (active; placebo; control) BOX

209. Intervention/exposure type (e.g., marine diet; plant diet; marine-based supplement; plant-based supplement; nuts; nut oil): BOX

210. Define the intervention/exposure components (e.g., ‘omega-3 rich diet;’ anchovy serving; fish oil supplement; flaxseed/linseed; rapeseed/canola; walnut oil; softgel capsule): BOX

211. If control/placebo, describe it (e.g., ‘omega-3 poor diet’) or what was used (e.g., olive oil, safflower oil): BOX

212. Specific source(s) of omega-3 fatty acids (i.e., type[s] of fish or plant): BOX

213. Per-dose/serving amount of omega-3 fatty acids (e.g., grams): BOX

214. Dose/serving frequency (e.g., once daily): BOX

215. Timing (e.g., with breakfast; any time): BOX

216. Route of administration: BOX

217. Total daily (mean; SD/SE; range) omega-3 fatty acid intake (e.g., grams): BOX

218. Type(s) of omega-3 fatty acid identified (i.e., DHA, EPA, DPA, ALA): BOX

219. Total daily amount/dose (mean; SD/SE; range) of each type of omega-3 fatty acid, or combination (e.g., EPA+DHA) (e.g., grams): BOX

220. Omega-3 fatty acid composition (%) of the exposure/intervention: BOX

221. Permitted or mandated amount/dose (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure) of omega-6 fatty acid intake (e.g., grams): BOX

222. Permitted or mandated omega-6/omega-3 ratio of the exposure/intervention: BOX

223. Other permitted or mandated type(s) of co-intervention (e.g., anti-oxidants such as Vitamin E; other dietary supplements; other foodstuff), including respective dose/servings (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure), with units: BOX

224. Total daily fat intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

225. Total daily caloric intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

226. Describe whether, and why, it was necessary or possible to change the standard asthma medications/treatments during the course of the study: BOX

227. If standard asthma medications/treatments were changed, then how? BOX

228. If there is an additional omega-3 fatty acid ‘active’ arm or, if none, then a control arm (e.g., placebo), type ARM5 in the text box (however, click here if there are no more arms/cohorts; this links directly to question #252) BOX

229. Re this study arm, what is the sample size at study entry? BOX

230. Sample size of those completing the study: BOX

231. Intervention length (e.g., weeks, months): BOX

232. Arm type (active; placebo; control) BOX

233. Intervention/exposure type (e.g., marine diet; plant diet; marine-based supplement; plant-based supplement; nuts; nut oil): BOX

234. Define the intervention/exposure components (e.g., ‘omega-3 rich diet;’ anchovy serving; fish oil supplement; flaxseed/linseed; rapeseed/canola; walnut oil; softgel capsule): BOX

235. If control/placebo, describe it (e.g., ‘omega-3 poor diet’) or what was used (e.g., olive oil, safflower oil): BOX

236. Specific source(s) of omega-3 fatty acids (i.e., type[s] of fish or plant): BOX

237. Per-dose/serving amount of omega-3 fatty acids (e.g., grams): BOX

238. Dose/serving frequency (e.g., once daily): BOX

239. Timing (e.g., with breakfast; any time): BOX

240. Route of administration: BOX

241. Total daily (mean; SD/SE; range) omega-3 fatty acid intake (e.g., grams): BOX

242. Type(s) of omega-3 fatty acid identified (i.e., DHA, EPA, DPA, ALA): BOX

243. Total daily amount/dose (mean; SD/SE; range) of each type of omega-3 fatty acid, or combination (e.g., EPA+DHA) (e.g., grams): BOX

244. Omega-3 fatty acid composition (%) of the exposure/intervention: BOX

245. Permitted or mandated amount/dose (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure) of omega-6 fatty acid intake (e.g., grams): BOX

246. Permitted or mandated omega-6/omega-3 ratio of the exposure/intervention: BOX

247. Other permitted or mandated type(s) of co-intervention (e.g., anti-oxidants such as Vitamin E; other dietary supplements; other foodstuff), including respective dose/servings (mean; SD/SE; range), frequency, method of delivery, and timing (relative to the omega-3 fatty acid exposure), with units: BOX

248. Total daily fat intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

249. Total daily caloric intake (mean; SD/SE; range) of the exposure/intervention, with units: BOX

250. Describe whether, and why, it was necessary or possible to change the standard asthma medications/treatments during the course of the study: BOX

251. If standard asthma medications/treatments were changed, then how? BOX

252. Was there a clear difference in daily total-gram omega-3 fatty acid intake across the study arms/cohorts/phases? YES CT NO N/A

253. Was there a clear difference in the respective daily intake amounts for each of the different omega-3 fatty acids across the study arms/cohorts/phases? YES CT NO N/A

254. Was there a clear difference in the total daily omega-6 fatty acid intake across the study arms/cohorts/phases? YES CT NO N/A

255. Was there a clear difference in the omega-6/omega-3 ratio of the daily fatty acid intake across the study arms/cohorts/phases? YES CT NO N/A

256. Was the daily overall fat intake equivalent across study arms/cohorts/phases? YES CT NO N/A

257. Was the daily caloric intake equivalent across study arms/cohorts/phases? YES CT NO N/A

258. Did control participants appear to receive more than trace amounts of omega-3 fatty acid? YES CT NO N/A

259. Were the standard medications/treatments for asthma equivalent or comparable across the study arms/cohorts/phases? YES CT NO N/A

260. Were the standard medications/treatments for concurrent conditions equivalent or comparable across the study arms/cohorts/phases? YES CT NO N/A

261. Identify any factors intentionally kept constant across the study (e.g., background diet for those in a study of the impact of supplements): BOX

262. Additional comments re the possible non-comparability of populations or interventions/exposures, and, other possible sources of bias: BOX

263. Any further comments about the study: BOX

264. Identify the question(s) this study addresses (select all that apply):

     • What is the evidence for the efficacy of omega-3 fatty acids to improve respiratory outcomes among individuals with asthma?

     • What is the evidence that the possible value (efficacy/association) of omega-3 fatty acids in improving respiratory outcomes is dependent on the: specific type of fatty acid (DHA, EPA, DPA, ALA, fish, fish oil); specific source (fish, plant, food, dietary supplement [fish oil, plant oil]); its serving size or dose (fish or dietary supplement); amount/dose of omega-6 fatty acids given as a co-intervention; ratio of omega-6/omega-3 fatty acids used; fatty acid content of blood lipid biomarkers; absolute fatty acid content of the baseline diet; relative fatty acid content of the baseline diet; tissue ratios of fatty acid (omega-6/omega-3) during the investigative period; intervention length; anti-oxidant use; and, the manufacturer and its product(s) (different purity; presence of other potentially active agents)?

     • What is the evidence that, in individuals with asthma, omega-3 fatty acids influence mediators of inflammation which are thought to be related to the pathogenesis of asthma?

     • Are omega-3 fatty acids effective in the primary prevention of asthma?

     • Among individuals with asthma, do omega-3 fatty acids alter the progression of asthma (i.e., secondary prevention)?

     • What is the evidence for adverse events, side effects, or counter-indications associated with omega-3 fatty acid use to treat or prevent asthma (DHA, EPA, DPA, ALA, fish oil, fish)?

     • What is the evidence that omega-3 fatty acids are associated with adverse events in specific subpopulations of asthmatic individual such as diabetics?

Quality Assessment Form—Randomized Controlled Trials

Quality Assessment Form—All Other Study Designs

Applicability Indices

For studies of the treatment and secondary prevention (i.e., to alter the progression) of asthma:

Assign ‘I’ to a study population of otherwise ‘healthy’ asthmatic North American individuals, namely those representing a somewhat broad demographic picture (i.e., gender, race), possibly exhibiting ‘typical’ conditions concomitant to asthma (e.g., atopy), living a ‘typical’ North American lifestyle (e.g., background diet), receiving ‘typical’ types and doses of asthma treatment (e.g., medications), yet without significant comorbid health problems.

Assign ‘II’ to a study population of asthmatic individuals representing a more restricted North American demographic picture (e.g., a North American sub-population), exhibiting more severe forms of ‘typical’ condition concomitant to asthma, living a less ‘typical’ North American lifestyle, receiving less ‘typical’ types and doses of asthma treatment, or having significant comorbid health problems (e.g., diabetes).

Assign ‘III’ to a study population of asthmatic individuals representing a narrow demographic picture that is not a ‘typical’ North American one (e.g., a population living outside North America), living a lifestyle that is not a ‘typical’ North American one (e.g., different background diet), exhibiting more severe forms of ‘typical’ condition concomitant to asthma, receiving less ‘typical’ types and doses of asthma treatment, or having significant comorbid health problems.

Assign ‘X’ when applicability cannot be ascertained due to incomplete reporting, particularly of the details defining the study population of asthmatic individual, including demographics, the diagnostic criteria and method to identify asthma, comorbid health problems, and, asthma care.

For primary prevention studies:

Assign ‘I’ to a study population of typical ‘healthy’ North Americans across a broad demographic spectrum, or, those otherwise ‘typical’ North Americans (i.e., no other significant health problems) across a broad demographic spectrum yet at risk to develop asthma (e.g., family history; environmental exposure to smoke; early history of respiratory infections).

Assign ‘II’ to a study population of North Americans, with or without the risk of developing asthma, yet representing a restricted demographic picture (e.g., a sub-population), lifestyle (e.g., different background diet), or other significant health problems (e.g., diabetes).

Assign ‘III’ to a study population living outside North America, with or without the risk of developing asthma, and thus representing --relative to North Americans-- a different demographic picture or lifestyle, or, other significant health problems (e.g., diabetes).

Assign ‘X’ when applicability cannot be ascertained due to incomplete reporting, particularly of the details defining the study population.