| Study | Quality grades assigned by reviewers (NS/TJ) | Confounders/variables analysed | Limitations/strengths | Findings |
| Preschool health outcomes | ||||
| Schluter et al,199819 | Sound/sound | SE, maternal factors, infant factors and selection factors | L: selection bias; recall bias. S: confounders analysis; sufficient numbers to achieve conclusive results | Increased risk among infants sleeping “away” from home: Infants “away” vs at home: OR 1.70 (1.09 to 2.66) |
| Stoneman et al, 199920 | Suboptimal/adequate | Race, sex, age, care-giver conflict, SE | L: Small highly selective group; volunteer sample—self-selection bias. S: Incorporation of both parental and teacher reports, consideration of child temperament | Increased sibling conflict, but not behaviour, depression or social isolation. With RM, low emotionality children experienced higher depression scores |
| Knudson-Cooper and Leuchtag, 198221 | Suboptimal/suboptimal | None | L: Selection bias. Confounders not analysed; no control group used | Children admitted to this unit have more moves than expected for the whole US population |
| School age health outcomes | ||||
| DeWit et al, 199822 | Sound/sound | Age, sex, location, family structure and resources (SE), parental control, self-esteem, peer drug use, neighbourhood “climate” | L: Respondent bias—based on self-report. S: Large dataset †with analysis of key confounders | ⩾3 Moves vs others: Indirect aggression, OR 1.26 (p,0.01) property offences OR 2.56, p<0.001; lifetime alcohol use OR 1.23, p<0.01; high attachment and lack of family conflict protective |
| Wood et al, 199323 | Sound/sound | SE, sex, race, family structure, parental employment status, parental education, maternal age, location | L: Only parental report used, reporting bias; sample demographics and attrition details lacking. S: Large national sample | No significant effect on development, growth, or learning disability rates seen. Frequent movers (>6 moves) more likely to have ⩾4 behavioural problems: OR 1.77 (1.37 to 2.29) |
| Simpson and Fowler, 199424 | Sound/Sound | Poverty, age, sex, race/ethnicity, region, mother’s marital status and education | L: Recall bias. School change and reason for move not assessed. S: Large representative sample with due attention to known confounders | Behavioural problems (⩾3 moves vs less) OR 2.3 (1.7 to 3.1); need of professional psychological help: OR 2.2 (1.4 to 2.0) |
| Verropoulou et al, 200225 | Sound/adequate | SE, parental education, family disruption, child’s age, sex | L: Attrition to study—potential systematic bias with loss of mobile families. S: Longitudinal design in an intergenerational study | RM not independently associated with aggression or anxiety |
| Beyers et al, 200326 | Adequate/adequate | SE, age, sex, ethnicity, family structure | L: Small numbers from each neighbourhood limits precision; selective attrition and predominance of European middle-income families limits generalisability. S: Longitudinal. Analysis of key confounders | Neighbourhood structure not independently related to externalising behaviour problems or change rates. Parental monitoring—associated with lower externalising behaviours in mobile neighbourhoods |
| Kelley et al, 200327 | Suboptimal/suboptimal | Child’s feelings about mother; maternal depression | L: Sample obtained by advertised invitation; self-selection bias. No adjustment for father’s rank in military. Highly selected sample, not generalisable. S: Use of validated measures | Time in residence (but not no of moves) related to outcomes, contributing to variance in loneliness (β coeff. = −0.31), peer relationships (β coeff. = −0.40) and self-worth (β coeff. = 0.23) |
L:, limitations; OR, odds ratio followed by 95% confidence interval in brackets; RM, residential mobility; S:, strengths; SE, socioeconomic factors or measures.