Table 3: Quality assessment of included studies using the NewCastle Ottawa Scale.

Study author, year, countrySelection Comparability of cases and controls on the basis of the design or analysisExposure Comment
Is the case definition adequate?Representativeness of the cases
Selection of controls
Definition of controls
Ascertainment of exposureThe same method of ascertainment of cases and controlsNonresponse rate

Carpenter et al. 2004, Europe [25]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants matched for age and the same survey area, randomly selected from the birth records (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, interviews (*)Yes (*)Described (*)Good quality

Blair et al. 1999, United Kingdom [22]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants born immediately before and after the index case (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, interviews (*)Yes (*)Described (*)Good quality

Arnestad et al. 2001, Norway [23]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants matched for sex and date of birth, randomly picked from the national register (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, mailed questionnaire (*)Yes (*)Described, 31% and 25% loss for the cases and controls, respectively (*)Good quality

Bubnaitiene et al. 2005, Lithuania [26]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants matched for date of birth and region, randomly picked (*)Yes, groups comparable. Not adjusted for most potential confounders (*)Yes, home visits in cases and mailed questionnaire in controls (*)No Described only for the cases, 22.2% loss (*)Good quality

Ruys et al. 2007, The Netherlands [28]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants who participated in another survey (*)No, groups not comparable. Adjusted for most potential confounders (*)Yes, home visits and direct interview in cases and direct interview in controls (*)Yes (*)Not describedGood quality

McGarvey et al. 2006, Ireland [27]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants matched for date of birth and population based area, randomly picked from birth register (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, home interview in both cases and controls (*)Yes (*)Described, 14% loss (*)Good quality

Fu et al. 2010, USA [30]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants matched for birth race, age, and birth weight, randomly picked from birth register (*)Yes, groups comparable. Not adjusted for most potential confounders (*)Yes, home visits and direct interview in cases and direct interview in controls (*)Yes (*)Described, 25% loss (*)Good quality

Klonoff-Cohen and Edelstein 1995, USA [19]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants matched for birth hospital, sex, race, and date of birth, randomly picked from birth register (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, telephonic interview in cases and controls (*)Yes (*)Described, 25% loss (*)Good quality

Blair et al. 2009/England [9]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, from the maternity database of hospital (*)No, groups not comparable. Not adjusted for most potential confounders Yes, home visits and questionnaire in cases and questionnaire in controls (*)NoDescribed, 5–14% loss in the two groups (*)Good quality

Vennemann et al. 2009, Germany [29]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, matched for age, gender, region, and sleep time (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, home visits and questionnaire in cases and controls (*)Yes (*)Described, 18–42% loss in the two groups (*)Good quality

L’Hoir et al. 1998, The Netherlands [21]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, matched for date of birth (*)No, groups not comparable. Not adjusted for most potential confounders Yes, home visits and questionnaire in cases and controls (*)Yes (*)Not describedGood quality

Brooke et al. 1997, Scotland [20]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, matched for age, season, and maternity unit (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, home visits and questionnaire in cases and controls (*)Yes (*)Described, ~25% loss (*)Good quality

Fleming 1996, EnglandYes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants born immediately before and after the index case (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, home visits and questionnaire in cases and controls (*)Yes (*)Described, ~9% loss (*)Good quality

Williams et al. 2002, New Zealand [24]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, randomly selected from all births, except home births (*)No, groups not comparable and also adjusted for most potential confounders (*)Yes, interview based in cases and controls (*)Yes (*)Described, 10–19% loss (*)Good quality

Ball 2003, United Kingdom [12]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, healthy infants and mothers, delivered at 36+ weeks. Followed up till 4 months of age (*)Yes, groups comparable. Not adjusted for most potential confounders (*)Yes, sleep logs were used to measure the exposure status (*)Yes (*)Described, ~40% loss (*)Good quality

McCoy et al. 2004, USA [14]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, followup of infants born at selected study hospitals (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, mailed questionnaire (*)Yes (*)Described, ~30% loss (*)Good quality

Lahr et al. 2007, USA [15]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, stratified sample drawn each month from recently filed birth certificates (*)Yes, groups comparable. Not adjusted for most potential confounders (*)Questionnaire based (*)Yes (*)Described, 26.5% loss (*)Good quality

Tan 2011, Malaysia [17]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, controls attending health facility (*)Yes, infants up to 6 months attending health clinics (*)Yes, groups comparable and also adjusted for most potential confounders (**)Face-to-face interviews using a pretested structured questionnaire (*)Yes (*)Described, <5% loss (*)Good quality

M llborg et al. 2011, Sweden [18]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, randomly selected families with infants who had reached 6 months of age (*)Yes, groups comparable and also adjusted for most potential confounders (**)Yes, mailed questionnaire (*)Yes (*)Described, 31.5% loss (*)
Good quality

Flick et al. 2001, USA [11]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, pregnant women enrolled at 28 wk, contacted at around 8 weeks after delivery (*)Yes, groups comparable. Not adjusted for most potential confounders (*)Yes, mailed questionnaire (*)Yes (*)Described, 2.2% loss (*)Good quality

Blair and Ball 2004, United Kingdom [13]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, healthy newborn infants and mothers at home (*)Yes, groups comparable. Not adjusted for most potential confounders (*)Yes, sleep logs and interviews (*)Yes (*)Described, <20% loss (*)
Good quality

Blair et al. 2010, United Kingdom [16]Yes, record validation (*)Yes, obviously representative series of cases (*)Yes, community controls (*)Yes, infants of all pregnant women residing in the 3 health districts of Avon; age group: birth to 4 years (*)Yes, groups comparable. Not adjusted for most potential confounders (*)Yes, mailed questionnaire (*)
Yes (*)Described, ~50% loss (*)
Good quality

One point, **two points.