Table 3.
Completeness of reporting on selected aspects of internal and external validity*
| Study | Preintervention | Postintervention/follow-up | Analysis | |||||||
| First author (reference) | Response rate at baseline | Baseline comparison of respondents and nonrespondents | Test for equivalence of study groups at baseline | Response rate at follow-up | Compare final sample and dropouts | Test for differential attrition by study group | Sample size determination (a priori statistical power analysis) | Intent-to-treat† | Unit of randomization or assignment | For nested samples: appropriate analysis/adjustment |
| Andersen (61) | Y | N | N | Y | N | N | N | N | Group (community) | Y |
| Barr (65) | NA | NA | Y | NA | NA | NA | N | Y | Individual | NA |
| Bodiya (66) | NA | NA | N | NA | NA | NA | N | Y | Individual | NA |
| Clark (72) | Y | N | N | N | N | Y | N | N | Individual | NA |
| Costanza (75) | Y | N | N | Y | Y | Y | N | N | Group (practices) | Y |
| Crane (58) | Y | N | Y | Y | Y | Y | N | N | Residences | NA |
| DeFrank (70) | Y | N | Y | Y | N | N | Y | N | Individual | NA |
| Drossaert (59) | NA | NA | Y | NA | NA | NA | N | Y | Individual | NA |
| Duan (64) | N | N | Y | Y | N | N | N | N | Group (church) | Y |
| Eaker (60) | Y | N | Y | Y | N | N | N | N | Group (county)‡ | N |
| Finney (23) | NA | NA | N | NA | NA | NA | Y | Y | Individual | NA |
| Goel (62) | NA | NA | Y | NA | NA | NA | N | Y | Individual | NA |
| Lipkus (32) | Y | N | Y | Y | N | N | N | N | Individual | NA |
| Margolis (73) | Y | N | Y | Y | N | Y | N | Y | Individual | NA |
| Mayer (67) | Y | N | Y | NA | NA | N | Y | Y | Individual | NA |
| Messina (76) | Y | N | Y | Y | Y | N | N | N | Individual | NA |
| Michielutte (74) | Y | Y | Y | Y | N | Y | N | N | Group (practices) | Y |
| Partin (63) | NA | NA | Y | NA | NA | NA | N | Y | Individual | NA |
| Quinley (71) | NA | NA | Y | NA | NA | NA | N | Y | Group (mammography facilities)‡ | N |
| Rakowski (68) | Y | N | Y | Y | N | N | Y | N | Individual | NA |
| Rimer (77) | Y | N | Y | Y | N | N | N | N | Individual | NA |
| Schapira (69) | NA | NA | Y | Y | N | Y | N | N | Individual | NA |
| Simon (31) | NA | NA | Y | NA | NA | NA | Y | Y | Individual | NA |
| Skinner (78) | Y | N | N | Y | N | Y | N | N | Individual | NA |
| Vernon (29) and del Junco (28) | Y | Y | Y | Y | Y | Y | Y | Y | Individual | NA |
*
Studies coded NA used a data source such as medical records to identify the study population, and the investigators were able to enumerate and track all eligible women; therefore, issues related to response rates, equivalence, and follow-up were not a problem. N = no; NA = not applicable; Y = yes.
†
We defined intent-to-treat analysis as including all eligible randomized persons in the outcome analysis whether or not they completed the intervention and/or all study surveys (79). For example, if participants were randomized to a study group but were later excluded from the outcome analysis because they did not complete follow-up surveys, a study was rated “N.”