Inferential Statistics - Essay Example

Research Hypothesis: We hypothesize that our new drug, hivaril, is more effective at raising t-cell counts in AIDs patients than traditional drug cocktails. Dependent Variables: T-cell counts Independent Variables: Drug treatment Selecting Subjects: Patients who have developed bona fide AIDs, not HIV, will be selected for this process. This will be done using national AIDs databases and hospital databases. Once 1000-1500 patients are randomly selected from among the nation, we will offer them participation in the study.

All drugs will be freely provided and there will be additional financial incentive. Patients with extremely low T-cell counts at imminent danger of dying will not be excluded but will be given special treatment as noted below in the discussion of ethical concerns. Study Design: Patients will be split into three groups: A placebo control group, a standard cocktail control group and a hivaril experimental group. For the hivaril group, in order to mask that a newer drug is being provided, the rest of the cocktail elements will be simulated by identical-looking placebos.

Patients will be monitored for six months regularly to check T-cell performance and other vectors of improved health, as well as to see for toxicity. Statistical Model: Our intent is to see, at a very high confidence level with a moderate margin of error (+/- 3% would be acceptable), if hivaril is comparable to or superior to both the cocktail control group and the placebo control group. The hypothesis is directional, with the null hypothesis being that hivaril is worse than the cocktail and identical to a placebo.

The drug would still be a major improvement if it were better than the placebo but slightly worse than the cocktail. The alternative hypothesis is as noted. Because three groups are being measured, and must be to preserve a placebo control, ANOVA will be used. Luckily, the sample size is large enough that no other modification is needed. Experimental Bias: Experimenters may be concerned about patient welfare enough to scuttle the study or invested in proving the efficacy of hivaril. Therefore, as is standard procedure, the study will be double-blind.

Selection Bias: It is possible that our selected subjects could have a salient shared characteristic aside from treatment. Therefore, before being placed into placebo control, cocktail control or hivaril experimental groupings, patients will be clustered based on race, age, current T-cell count, gender, social status and other demographic variables. These will be distributed evenly and the final results will break down the results along demographic lines as well as aggregated. Ethical Concerns: When treating extremely ill people with an experimental treatment, there is always a serious ethical dilemma.

If we honestly believe hivaril is equal to a more dangerous cocktail, isn't it irresponsible not just to give it? They are already so sick that the risks are outweighed by the benefits. On the other hand, since knowledge is not complete on hivaril's efficacy, isn't it grossly irresponsible to give people an inferior intervention? Thus, we are classifying people into high and low performers. High performers are treated normally, but low performers will be specially monitored, with extensive consultation with a primary physician.

They will be given especial attention in explaining their experimental obligations. They will not be prescribed placebos: While a placebo is useful, cocktail drugs have already been measured against a placebo, so their inclusion in the study is strictly formal quality control. Any substantial decline, irrespective of cause, will cause them to be placed onto the established cocktail and informed of the drug they were on, with early termination taken into account in study data. They will also be given the option to continue taking hivaril.