Assignment 1 - Biostatistics
STUDY DESIGNS AND FLAWS (40 pts.)

For problems 1 - 4 answer any specific questions asked about the study in part (a) and then for part (b) for each problem do the following: Indicate whether sound experimental designs were used and describe any design problems that are evident. Also discuss biases or confounding that might have arisen from design deficiencies. Essential information about the design is omitted in some of the study descriptions, and this should be noted. Such omissions are all too common in the medical and health care literature.

1.“This prospective study was designed to evaluate the efficacy of cisplatin in treating cancer-associated hypercalcemia in humans. Thirteen patients with severe hypercalcemia refractory to rehydration were treated with a 24-hour infusion of cisplatin, 100 mg/m2. Serial measurements of serum calcium and tumor size were made following cisplatin treatment and compared with pretreatment values. Nine patients (69%) achieved normocalcemia after treatment with cisplatin; and mean [average] duration of benefit was 38 days in these patients… We conclude that cisplatin can control malignant hypercalcemia for relatively long periods.”

Taken from Lad TE, Mishoulam HM, Shevrin DH, et al “Treatment of cancer-associated hypercalcemia with cisplatin.” Arch Intern Med 147: 329-332, 1987.

a) What does the term “serial” mean at the beginning of the third sentence? (1 pt.)

b) Discuss any design deficiencies (see above). (3 pts.)

2. “The purpose of this study was to compare the effect of three methods of chest tube management on mediastinal drainage volume in cardiac surgery patients: (1) a standard single-lumen thoracic catheter with suction of -20 cm H2O and intermittent chest tube stripping (standard method), (2) the standard method with addition of venting poststripping pressures over -20 cm H2O (venting method), and (3) use of a sump catheter with no stripping (sump method)… Thirty men undergoing closed mediastinal drainage after first-time cardiac surgery were studied, 10 in each group… Data was collected first for the standard method because it created the least disruption in usual agency practice, second for the venting method, and last for the sump method, which represented the greatest change from usual routines.”

Taken from Duncan CR, Erickson RS, Weigel RM, “Effect of chest tube management on drainage after cardiac surgery”, Heart Lung 16:1-9, 1987.

a) What method of statistical analyses could be used to compare the drainage rates between the three methods? Justify your choices. (3 pts.)

b) Discuss any design deficiencies (see above). (3 pts.)

3. “The first [method] (experimental group) was based on a new approach which claimed that post-partum breast discomfort in non-breastfeeding women would be reduced if milk was extracted from the breasts. The second approach (hospital policy group) involved adhering to the normal hospital routine which was a combination of traditional methods. … Non-breastfeeding women admitted to Ward A were assigned to the experimental group (nA= 95), and non-breastfeeding women admitted to Ward B acted as the control or hospital policy group (nB = 57). … Three times daily, participants entered details on the questionnaire relating to self-assessment of pain, use of analgesia and methods employed in supressing lactation”.

Taken from Webster J, “Lactation suppression: A pilot study” Australian Journal of Advanced Nursing, 4:36 – 40, 1986.

a) What method of statistical analyses might be used in analyzing data from this study?
Justify your choices.(2pts.)

b) Discuss any design deficiencies (see above). (3 pts.)

4. “This experimental design study as instituted with two sample groups of a population of traumatic head-injured patients utilizing two health care facilities… The first sample consisted of 15 patients who were admitted to a setting that employed a controlled structured sensory-simulation program for head-injured patients. This was the experimental group. The control group consisted of 15 head-injured patients who received nursing care that did not include planned, structured sensory stimulation… The patients in the control group, were matched on a one-to-one basis with patients in the experimental group on the basis of sex, age range, approximate type of injury, Glasgow Coma Scale Score, and length of time post injury… Scores of cognitive function in the experimental group were compared to those of the control group.”

Taken from Kater KM, “Response of head-injured patients to sensory stimulation”, West Journal of Nursing Research, 11: 20 – 33, 1989.

a) What methods of statistical analyses could be used to analyze these data? Justify your
choices. (2 pts.)

b) Discuss and design deficiencies. (2 pts.)

5. Researchers wish to compare the hospital costs associated with the treatment of and the 30-day mortality of heart attack patients who have undergone right heart catheterization (i.e. Swan-Ganz catheterization) to those patients who do have the procedure. Swan-Ganz catheters are used to monitor the following:

  • pulmonary artery systolic pressure
  • pulmonary artery mean pressure
  • pulmonary diastolic pressure
  • pulmonary capillary wedge pressure
  • cardiac index
  • right atrial pressure

How might researchers conduct such a study? Be thorough in your description of how the study might be carried out and what variables will be measured on each patient. What types of statistical analyses might be used for the data gathered? Justify your choices. (15 pts. total, see below for breakdown)

a) Study design. (5 pts.)

b) Variables to measure (be thorough but I realize your not M.D.’s) (5 pts.)

c) Statistical analyses you might use. (5 pts.)

6. Explain why a cancer patient involved in a 2 yr. clinical trial that survives for 7 months and then gets hit by a bus and dies still provides useful information about the effectiveness of the medical treatment he/she was receiving. (2 pts.)

7. In a retrospective study of alcohol use and lung cancer, researchers found that individuals who are drink had twice the risk of dying of lung cancer than non-drinkers. Researchers concluded that “drinking alcohol causes lung cancer”. Comment on this conclusion. (3 pts.)

8. What is the fundamental difference between and observational vs. experimental studies? (1 pt.)