Confounding variables control

Another occupational study compared 53 lead-exposed male workers (2 nonwhite, 51 white) (mean PbB, 47.4 pg/dL range, 44-51 pg/dL) from a plant processing lead and cadmium compounds with a control group of 52 workers (8 nonwhite, 44 white) (mean PbB, 8.1 pg/dL, with none exceeding 20 pg/dL) from a nonlead industry (de Kort et al. 1987). Blood pressure levels were positively correlated with PbB and urine cadmium levels, but not with blood cadmium levels. The correlation for systolic blood pressure and PbB level remained significant after controlling for confounding variables. 

These three studies indicate that exposure to low levels of lead may impact negatively upon children s hearing. However, the authors of the Robinson study did not state whether age and other possible confounding variables were controlled for. Similarly, in the NHANES study, age may have been a confounding variable. 

Studies to date have demonstrated an extremely weak correlation (if any) between photochemical-oxidant pollution and hospital admissions. If such correlations do exist, observations must be carried out over considerably longer periods with adequate provisions for controlling confounding variables, such as meteorologic factors. 

Random assignment is the most important element of a controlled trial. Without it, patients most likely to respond could be preferentially assigned to one treatment arm, and any difference in efficacy would be secondary to this bias. When random assignment is used, a variety of confounding variables are equalized across the groups, including those of which the investigators are unaware ( 2). 

There is other evidence that transdermal estrogen replacement therapy has relatively little effect on hemostasis. In a case control study, 155 consecutive patients with a first documented episode of idiopathic venous thromboembolism, 92 of whom had had a pulmonary embolism and 63 a deep venous thrombosis, were compared with 381 healthy matched controls (88). Overall, 32 (21%) of the cases and 27 (7%) of the controls were current users of oral estrogen replacement therapy, whereas 30 (19%) cases and 93 (24%) controls were current users of transdermal estrogen replacement therapy. After adjustment for potential confounding variables, the odds ratios for venous thromboembolism in current users of oral and transdermal estrogen replacement therapy compared with non-users were 3.5 (95% Cl = 1.8, 6.8) and 0.9 (0.5, 1.6) respectively. Estimated risk for venous thromboembolism in current users of oral estrogen replacement therapy compared with transdermal users was 4.0 (1.9, 8.3). 

In a case-control study 155 postmenopausal women who had had venous thromboembolism were compared with 381 matched controls (91). In all, 32 cases and 27 controls were current users of oral replacement therapy, whereas 30 cases and 93 controls were current users of transdermal products. After adjustment for potential confounding variables, the estimated risk ratio for venous thromboembolism in current users of the oral products compared with the transdermal users was 4.0 (1.9-8.3). This is strong evidence that the transdermal route was considerably safer. However, the conclusions of different studies continue to conflict with one another, no doubt in part because of variations in the formulations and patterns of use of the products. 

It is important to minimize or control for the impact of confounding variables in any study. In order to do so, it is imperative that all possible variables that may have an effect on the primary outcome be identified. These usually include demographic variables such as age, sex, income level, education level, and ethnicity. Health-related variables such as comorbidities and severity of illness should also be recorded. Variables also may be identified that cannot be measured. These may include outside education, changes in family structure or support system, and drastic changes in health status not related to the pertinent disease state. 

These increase or decrease systematically as the independent variable increases or decreases. Their effects are known as systematic variation. This form of variation can be disentangled from that caused directly by treatments by incorporating appropriate controls in the experiment. A control is really just another treatment where a potentially confounding variable is adjusted so that its effects, if any, can be taken into account. The results from a control may therefore allow an alternative hypothesis to be rejected. There are often many potential controls for any experiment. 

Incorporating appropriate controls to show that potential confounding variables have little or no effect. 

Observational cohort or case control designs have been used to determine ADR relationships with specific therapeutic classes. " Medical claims data are often used in these studies and caution should be warranted due to lack of definite confirmation of drug exposure and the potential for confounding variables. However, these studies have been beneficial in determining risk of ADRs with specific classes (e.g., NSAIDs and the risk of peptic ulcer disease). 

---