Table 2.
Tests of heterogeneity, odds ratios (ORs), and 95% confidence intervals (CIs) for potential explanatory variables of repeat mammography screening*
| Study variable | No. of estimates | First author (reference) | Q | df | P | I 2 | OR (95% CI) |
| All studies | 27† | 69.5 | 26 | <.001 | 62.6 | 1.39 (1.27 to 1.52) | |
| Age, y | |||||||
| ≥40 | 9 | DeFrank (70), Eaker (60), Finney (23), Goel (62), Margolis (73), Partin (63), Rimer (77), Schapira (69), Simon (31) | 21.9 | 8 | .005 | 63.5 | 1.31 (1.08 to 1.59) |
| ≥50 | 18 | Andersen (61), Barr (65), Bodiya (66), Clark-1 (72), Clark-2 (72), Costanza-1 (75), Costanza-2 (75), Crane (58), Drossaert (59), Duan (64), Lipkus (32), Mayer (67), Messina (76), Michielutte (74), Quinley (71), Rakowski (68), Skinner (78), Vernon (29) | 44.8 | 17 | <.001 | 62.1 | 1.44 (1.29 to 1.60) |
| Study setting | |||||||
| Health care | 19 | Barr (65), Bodiya (66), Clark-1 (72), Clark-2 (72), Costanza-1 (75), Costanza-2 (75), DeFrank (70), Finney (23), Lipkus (32), Margolis (73), Mayer (67), Messina (76), Michielutte (74), Quinley (71), Rakowski (68), Rimer (77), Schapira (69), Simon (31), Skinner (78) | 59.9 | 17 | <.001 | 71.6 | 1.40 (1.20 to 1.65) |
| Community | 8 | Andersen (61), Crane (58), Drossaert (59), Duan (64), Eaker (60), Goel (62), Partin (63), Vernon (29) | 9.3 | 7 | .230 | 25.0 | 1.36 (1.22 to 1.50) |
| Screening interval | |||||||
| 1 year | 18 | Bodiya (66), Clark-1 (72), Clark-2 (72), DeFrank (70), Duan (64), Finney (23), Goel (62), Lipkus (32), Mayer (67), Michielutte (74), Partin (63), Quinley (71), Rakowski (68), Rimer (77), Schapira (69), Simon (31), Skinner (78), Vernon (29) | 48.3 | 17 | <.001 | 64.8 | 1.41 (1.27 to 1.57) |
| 2 years | 7 | Andersen (61), Barr (65), Costanza-1 (75), Costanza-2 (75), Crane (58), Drossaert (59), Messina (76) | 14.6 | 6 | .023 | 59.0 | 1.44 (1.16 to 1.78) |
| Age dependent | 2 | Eaker (60), Margolis (73) | NA | NA | NA | NA | NA |
| Study design | |||||||
| Design 1: measured one pre- and one postintervention mammogram | 20 | Andersen (61), Barr (65), Bodiya (66), Clark-1 (72), Costanza-1 (75), Crane (58), DeFrank (70), Drossaert (59), Duan (64), Eaker (60), Finney (23), Goel (62), Margolis (73), Mayer (67), Michielutte (74), Partin (63), Quinley (71), Rakowski (68), Schapira (69), Simon (31) | 60.9 | 19 | <.001 | 68.8 | 1.44 (1.29 to 1.62) |
| Design 2: measured two postintervention mammograms | 7 | Clark-2 (72), Costanza-2 (75), Lipkus (32), Messina (76), Rimer (77), Skinner (78), Vernon (29) | 6.64 | 6 | .355 | 9.7 | 1.26 (1.12 to 1.42) |
| Mammography data source | |||||||
| Medical records, administrative, or program data | 14 | Barr (65), Bodiya (66), Clark-1 (72), Clark-2 (72), Drossaert (59), Finney (23), Goel (62), Mayer (67), Michielutte (74), Partin (63), Quinley (71), Rakowski (68), Simon (31), Skinner (78), | 43.7 | 13 | <.001 | 70.3 | 1.42 (1.25 to 1.62) |
| Self-report | 10 | Andersen (61), Costanza-1 (75), Costanza-2 (75), Crane (58), Duan (64), Eaker (60), Lipkus (32), Messina (76), Rimer (77), Vernon (29) | 9.8 | 9 | .367 | 8.2 | 1.36 (1.23 to 1.52) |
| Medical record or self-report | 3 | DeFrank (70), Margolis (73), Schapira (69) | NA | NA | NA | NA | NA |
| Intervention strategy class 1‡ | |||||||
| Reminder | 8 | Barr (65), Bodiya (66), Finney (23), Goel (62), Mayer (67), Quinley (71), Schapira (69), Simon (31) | 42.2 | 7 | <.001 | 83.4 | 1.79 (1.41 to 2.29) |
| Education/motivation | 7 | Clark-1 (72), Clark-2 (72), DeFrank (70), Drossaert (59), Eaker (60), Partin (63), Vernon (29) | 2.5 | 6 | .868 | 0.0 | 1.25 (1.14 to 1.38) |
| Counseling | 12 | Andersen (61), Costanza-1 (75), Costanza-2 (75), Crane (58), Duan (64), Lipkus (32), Margolis (73), Messina (76), Michielutte (74), Rakowski (68), Rimer (77), Skinner (78) | 18.6 | 11 | .068 | 40.9 | 1.28 (1.15 to 1.43) |
| Intervention strategy class 2 | |||||||
| Barriers-specific telephone counseling | |||||||
| Yes | 9 | Andersen (61), Costanza-1 (75), Costanza-2 (75), Crane (58), Lipkus (32), Messina (76), Michielutte (74), Rakowski (68), Rimer (77) | 10.1 | 8 | .261 | 20.5 | 1.35 (1.19 to 1.53) |
| No | 18 | Barr (65), Bodiya (66), Clark-1 (72), Clark-2 (72), DeFrank (70), Drossaert (59), Duan (64), Eaker (60), Finney (23), Goel (62), Margolis (73), Mayer (67), Partin (63), Quinley (71), Schapira (69), Simon (31), Skinner (78), Vernon (29) | 59.4 | 17 | <.001 | 71.4 | 1.40 (1.25 to 1.58) |
| Intervention strategy class 3 | |||||||
| Single | 14 | Barr (65), Bodiya (66), Clark-1 (72), Clark-2 (72), Goel (62), Drossaert (59), Duan (64), Finney (23), Margolis (73), Mayer (67), Quinley (71), Schapira (69), Simon (31), Vernon (29) | 51.6 | 13 | <.001 | 74.8 | 1.50 (1.30 to 1.73) |
| Multiple | 13 | Andersen (61), Costanza-1 (75), Costanza-2 (75), Crane (58), DeFrank (70), Eaker (60), Lipkus (32), Messina (76), Michielutte (74), Partin (63), Rakowski (68), Rimer (77), Skinner (78) | 15.1 | 12 | .237 | 20.5 | 1.28 (1.17 to 1.39) |
| Delivery mode class 1§ | |||||||
| Mail only | 10 | Clark-1 (72), Clark-2 (72), DeFrank (70), Drossaert (59), Finney (23), Mayer (67), Partin (63), Quinley (71), Simon (31), Vernon (29) | 26.1 | 9 | .002 | 65.5 | 1.34 (1.19 to 1.51) |
| Telephone only | 3 | Barr (65), Duan (64), Lipkus (32) | NA | NA | NA | NA | NA |
| Mail plus telephone | 10 | Bodiya (66), Costanza-1 (75), Costanza-2 (75), Crane (58), Goel (62), Messina (76), Michielutte (74), Rakowski (68), Rimer (77), Skinner (78) | 20.9 | 9 | .013 | 56.8 | 1.41 (1.16 to 1.71) |
| Mail plus in person | 2 | Margolis (73), Schapira (69) | NA | NA | NA | NA | NA |
| Community education plus other modes | 2 | Andersen (61), Eaker (60) | NA | NA | NA | NA | NA |
| Delivery mode class 2 | |||||||
| Single | 13 | Barr (65), Clark-1 (72), Clark-2 (72), DeFrank (70), Drossaert (59), Duan (64), Finney (23), Lipkus (32), Mayer (67), Partin (63), Quinley (71), Simon (31), Vernon (29) | 35.5 | 12 | <.001 | 66.2 | 1.41 (1.26 to 1.58) |
| Multiple | 14 | Andersen (61), Bodiya (66), Costanza-1 (75), Costanza-2 (75), Crane (58), Eaker (60), Goel (62), Margolis (73), Messina (76), Michielutte (74), Rakowski (68), Rimer (77), Schapira (69), Skinner (78) | 33.8 | 13 | <.001 | 61.5 | 1.38 (1.16 to 1.64) |
| Control group type | |||||||
| Active | 13 | Barr (65), Costanza-1 (75), Costanza-2 (75), DeFrank (70), Drossaert (59), Finney (23), Goel (62), Lipkus (32), Michielutte (74), Rakowski (68), Rimer (77), Schapira (69), Simon (31) | 35.0 | 12 | <.001 | 65.7 | 1.40 (1.19 to 1.64) |
| Survey only | 12 | Andersen (61), Clark-1 (72), Clark-2 (72), Crane (58), Duan (64), Eaker (60), Margolis (73), Mayer (67), Messina (76), Partin (63), Skinner (78), Vernon (29) | 29.6 | 11 | .002 | 62.8 | 1.36 (1.16 to 1.59) |
| No contact | 2 | Bodiya (66), Quinley (71) | NA | NA | NA | NA | NA |
| Use of theoretic framework | |||||||
| Yes | 19 | Andersen (61), Barr (65), Clark-1 (72), Clark-2 (72), Costanza-1 (75), Costanza-2 (75), Crane (58), DeFrank (70), Drossaert (59), Duan (64), Finney (23), Lipkus (32), Mayer (67), Messina (76), Michielutte (74), Rakowski (68), Rimer (77), Skinner (78), Vernon (29) | 47.7 | 18 | <.001 | 62.2 | 1.36 (1.21 to 1.53) |
| No | 8 | Bodiya (66), Eaker (60), Goel (62), Margolis (73), Partin (63), Quinley (71), Schapira (69), Simon (31) | 21.8 | 7 | .003 | 67.9 | 1.49 (1.22 to 1.81) |
| Theoretic construct | |||||||
| Barriers | |||||||
| Yes | 14 | Andersen (61), Barr (65), Costanza-1 (75), Costanza-2 (75), Crane (58), DeFrank (70), Duan (64), Lipkus (32), Messina (76), Michielutte (74), Rakowski (68), Rimer (77), Skinner (78), Vernon (29) | 25.2 | 13 | .022 | 48.3 | 1.35 (1.19 to 1.52) |
| No | 13 | Bodiya (66), Clark-1 (72), Clark-2 (72), Drossaert (59), Eaker (60), Finney (23), Goel (62), Margolis (73), Mayer (67), Partin (63), Quinley (71), Schapira (69), Simon (31) | 44.1 | 12 | <.001 | 72.8 | 1.45 (1.25 to 1.68) |
| Stage of change | |||||||
| Yes | 10 | Clark-1 (72), Clark-2 (72), Crane (58), Lipkus (32), Messina (76), Michielutte (74), Rakowski (68), Rimer (77), Skinner (78), Vernon (29) | 9.7 | 9 | .376 | 7.2 | 1.31 (1.18 to 1.45) |
| No | 17 | Andersen (61), Barr (65), Bodiya (66), Costanza-1 (75), Costanza-2 (75), DeFrank (70), Drossaert (59), Duan (64), Eaker (60), Finney (23), Goel (62), Margolis (73), Mayer (67), Partin (63), Quinley (71), Schapira (69), Simon (31) | 59.0 | 16 | <.001 | 72.9 | 1.44 (1.27 to 1.64) |
| Use of tailoring | |||||||
| Yes | 14 | Andersen (61), Clark-1 (72), Clark-2 (72), Costanza-1 (75), Costanza-2 (75), Crane (58), Duan (64), Lipkus (32), Messina (76), Michielutte (74), Rakowski (68), Rimer (77), Skinner (78), Vernon (29) | 14.7 | 13 | .325 | 11.7 | 1.32 (1.21 to 1.44) |
| No | 13 | Barr (65), Bodiya (66), DeFrank (70), Drossaert (59), Eaker (60), Finney (23), Goel (62), Margolis (73), Mayer (67), Partin (63), Quinley (71), Schapira (69), Simon (31) | 54.1 | 12 | <.001 | 77.8 | 1.46 (1.26 to 1.70) |
*
Odds ratios and 95% confidence intervals were computed under random-effects model assumptions for heterogeneous subgroups and under fixed-effects model assumptions for homogeneous subgroups. A statistically significant Q indicates a heterogeneous distribution of study effect sizes, which may then warrant additional subgroup analyses (30). Statistical tests were two-sided. Categories of a variable were rated NA if there were fewer than five estimates; heterogeneity tests were not performed for these categories. I2 = percentage of total variation across studies that is because of heterogeneity rather than chance (33); NA = not applicable; Q = the Cochran Q statistic.
†
Two of the 25 studies, Clark et al. (72) and Costanza et al. (75), provided two estimates, so the total number of estimates was 27. We designated these estimates 1 and 2.
‡
Some studies used more than one intervention strategy, for example, reminder plus counseling. We classified those studies based on the most intensive strategy used. Intervention strategies were classified and analyzed in three different ways.
§
Delivery modes were classified and analyzed in two different ways.