Table 2. Effects of the interventions on the primary outcome compared with control, by stratum.
| Outcome | Study arm and stratum | Clusters | Individual-level prevalence | Cluster-level prevalence | Crude risk difference | Stratified risk difference † | F test † |
|---|---|---|---|---|---|---|---|
| n | n/N (%) | Mean (SD) | RD (95% CI) | RD (95% CI) | p-value | ||
| Febrile patients tested for malaria | Control | 12 | 432/1536 (28%) | 33.9% (29) | 0 | 0 | 0.47 |
| Enugu | 7 | 339/1336 (25%) | 24.8% (19) | ||||
| Udi | 5 | 93/200 (47%) | 46.7% (38) | ||||
| Provider | 14 | 416/1832 (23%) | 48.3% (33) | 14.4 (-11.2, 40.0) | 14.2 (-11.0, 39.4) | ||
| Enugu | 8 | 287/1463 (20%) | 50.0% (39) | ||||
| Udi | 6 | 129/369 (35%) | 46.0% (28) | ||||
| Provider-school | 14 | 231/1496 (15%) | 36.5% (33) | 2.5 (-23.0, 28.1) | 2.4 (-22.8, 27.6) | ||
| Enugu | 8 | 133/1266 (11%) | 24.8% (22) | ||||
| Udi | 6 | 98/230 (43%) | 52.0% (41) | ||||
| Test positive patients receiving ACT | Control | 12 | 238/320 (74%) | 64.2% (28) | 0 | 0 | 0.76 |
| Enugu | 7 | 200/265 (75%) | 71.2% (25) | ||||
| Udi | 5 | 38/55 (69%) | 54.3% (32) | ||||
| Provider | 14 | 95/142 (67%) | 56.1% (31) | -8.1 (-33.3, 17.0) | -6.9 (-29.6, 15.8) | ||
| Enugu | 8 | 6/87 (79%) | 73.8% (19) | ||||
| Udi | 6 | 26/55 (47%) | 35.3% (31) | ||||
| Provider-school | 12 | 59/98 (60%) | 56.4% (33) | -7.8 (-33.5, 17.9) | -7.8 (-30.9, 15.4) | ||
| Enugu | 7 | 46/68 (68%) | 67.5% (26) | ||||
| Udi | 5 | 13/30 (43%) | 41.0% (38) | ||||
| Test negative patients receiving an antimalarial | Control | 11 | 51/112 (46%) | 55.6% (38) | 0 | 0 | 0.28 |
| Enugu | 6 | 34/74 (46%) | 60.8% (40) | ||||
| Udi | 5 | 17/38 (45%) | 49.4% (40) | ||||
| Provider | 14 | 94/274 (34%) | 30.9% (33) | -24.7 (-55.5, 6.2) | -24.9 (-56.0, 6.2) | ||
| Enugu | 8 | 70/200 (35%) | 24.3% (28) | ||||
| Udi | 6 | 24/74 (32%) | 39.7% (38) | ||||
| Provider-school | 13 | 61/133 (46%) | 43.7% (42) | -12.0 (-43.3, 19.4) | -12.5 (-44.2, 19.1) | ||
| Enugu | 8 | 48/65 (74%) | 56.1% (48) | ||||
| Udi | 5 | 13/68 (19%) | 23.7% (22) | ||||
| Treatment according to malaria guidelines ‡ | Control | 12 | 299/1536 (20%) | 22.6% (23) | 0 | 0 | 0.36 |
| Enugu | 7 | 240/1336 (18%) | 17.6% (15) | ||||
| Udi | 5 | 59/200 (30%) | 29.6% (31) | ||||
| Provider § | 14 | 275/1832 (15%) | 36.4% (32) | 13.8% (-8.0, 35.7) | 13.8% (-8.3, 35.8) | ||
| Enugu | 8 | 199/1463 (14%) | 43.5% (38) | ||||
| Udi | 6 | 76/369 (21%) | 27.0% (21) | ||||
| Provider-school § | 14 | 131/1496 (9%) | 23.5% (26) | 1.0% (-20.9, 22.9) | 0.9% (-21.1, 22.9) | ||
| Enugu | 8 | 63/1266 (5%) | 13.0% (14) | ||||
| Udi | 6 | 68/230 (30%) | 37.6% (32) |
TABLE NOTES
ACT = Artemisinin-based Combination Therapy. Only those with complete data for all components of the primary outcome are included. Hence, for example, 1,079 people are shown as having been tested, 58 fewer than the 1,137 in Fig 3, due to the 11+29+18 missing values noted there.
†Stratified analysis of cluster-level summary measures, F test of the null hypothesis of no differences between the three treatment arms, therefore 2 numerator degrees of freedom.
‡The between-cluster coefficient of variation was 0.26 in the Provider arm, and 0.19 in the Provider-school arm. A further comparison between arms adjusted for: facility type, stock-out of ACTs in past 4 weeks, age and sex of patient, tertile of socio-economic status from principal component analysis, whether the patient had previously sought treatment, and whether they asked for a blood test. Missing values in these variables forced the omission of 577 people. Compared to control, this yielded a benefit of Provider of 8.4% (95% CI: -4.4 to 21.3%) and of Provider-school of 7.1% (95% CI: -5.7 to 20%).
§The stratified analysis of Provider-school arm versus Provider arm showed a difference of -12.9% (95% CI -34.0 to 8.3%).