Means (standard deviations) by manufacturer, Jones n = 25, Smith n = 34.
| Variable | Jones mean (SD) |
Smith mean (SD) |
Test Results t(df) = test stat, p value |
| Age in years | 42.5 (14.2) | 39.6 (15.1) | t(57) = 0.74, p = .46 |
| Dose in gm/M2 | 22.58 (6.95) | 23.84 (8.57) | t(57) = 0.60, p = .55 |
| Serum creatinine in mg/dl | 1.01 (0.34) | 1.02 (0.29) | t(57) = 0.08, p = .94 |
| Weight in kg | 74.1 (16.8) | 82.5 (22.4) | t(57) = 1.58, p = .12 |
No significant differences were found.
| Variable | Jones cases / n (%) |
Smith cases / n (%) |
Yates Chi-square statistics |
| Sex - % men | 11 / 25 (44%) | 22 / 34 (65%) | c2(1, N = 59) = 1.74, p = .19 |
| Diag - % with leukemia | 19 / 25 (76%) | 23 / 34 (68%) | c2(1, N = 59) = 0.17, p = .68 |
| Stage - % in relapse | 10 / 25 (40%) | 17 / 24 (50%) | c2(1, N = 59) = 0.25, p = .62 |
| Toxicity - % w/ toxicity | 11 / 25 (44%) | 3 / 34 (9%) | c2(1, N = 59) = 8.00, p = .0047 |
The only significant association is with toxicity. See Jolson, 1992, p. 503, Table 3 for the published version of this table.
| Risk Factor | cRR^ (95% confidence interval for RR) |
| Age of at least 50 | 3.6 (95% CI: 1.5, 8.7) |
| Sex male | 0.8 (95% CI: 0.3, 2.0) |
| Diagnosis Leukemia | 5.3 (95% CI: 0.8, 37.2) |
| Stage Relapse | 7.1 (95% CI: 1.7, 29.0) |
| Manuf Jones | 5.0 (95% CI: 1.6, 16.0) |
AGE, STAGE, and MANUF are strong risk factors for TOX. DIAG is associated with TOX, but not significantly so. SEX does not appear to be a risk factor.
For an extraneous factor to confound the relationship between MANUF and TOX, it must be associated with both. Since AGE, DOSE, SCR, WEIGHT, SEX, DIAG, and STAGE do not appear to be associated with the MANUF (see Analysis 1), we do not expect confounding to occur. This analysis is only preliminary. A better assessment of the potential for confounding occurs in the next analysis.
Adjusted relative risks of MANUF for TOX, various adjustment factors.
| Adjustment (Stratification) Factor | aRR^ |
| Age (>= 50) | 4.9 |
| Sex (Male) | 5.9 |
| Diagnosis (Leukemia) | 4.5 |
| Stage (Relapse) | 6.1 |
See Jolson, 1992, p. 503, Table 5 for a similar type of analysis and commentary.
Since the crude relative risk and adjusted relative risks all fall between 4 and 6, the potential for confounding is small. We can therefore report the crude relative risk as an unbiased measure of association: cRR^ = 5.0 (95% CI:1.6, 16.0). A test of significance shows: Chi-square, Yates(1, N = 59) = 8.00, p = .0048. Therefore, there is a significant increase in risk of toxicity associated with the Jones product (5x the risk, on average). This estimate is imprecise, because of the small number of cases.