Question 1

We will be conducting a meta-analysis to investigate whether surgery (cystectomy) combined with radiotherapy has a survival benefit (measured at 3-years) compared to radiotherapy alone for patients with bladder cancer.

a) Conduct a keyword search in (ovid) Medline (untick -Map Term to Subject Heading') using the following search terms:

bladder carcinoma

radical radiotherapy

preoperative radiotherapy

radical cystectomy

controlled trial

Combine these with an 'and' command. Present your search results in a table showing the number of studies found at each step.

b) Your search above should have found one study only. Extract the data from this study to be included in the meta-analysis. Hint we want to look at 3 year survival comparing the two arms of the trial.

c) There were two other studies found that investigate three year survival, their results are shown in the following table:

3 year survival                                          Cystectomy                     Radiotherapy

                                                                 Survived          Total        Survived       Total

Miller 1977                                                 18                     35             7                    32

Sell 1991                                                    39                     88             31                  95

Calculate the study specific odds ratios comparing survival for cystectomy with survival for radiotherapy. Then calculate the Mantel-Haenszel pooled odds ratio.

d) Further analysis finds an I-squared =0.0%. Is your analysis in part c) appropriate?

e) Interpret your pooled odds ratio from part c). A test was conducted to see if the pooled odds ratio is significantly different from 1 and this test has a p-value = 0.0011, use this information to help with your interpretation.

Question 2

Figure 1 A directed acyclic graph (DAG) for the causal effect of malnutrition on mortality.

a) What is the primary relationship of interest in the DAG?

b) How many confounders are shown for this primary relationship of interest?

c) Does socioeconomic status need to be accounted for in any models that investigate the primary relationship of interest based on the DAG?

d) How many different paths does HIV status have to the mortality outcome?

e) What was the primary reason for use of the DAG in the paper?

Question 3

a) Explain a 'component Cause using the Rothman framework In your answer you need to mention how sufficient and necessary causes relate to a component cause

b) Is this statement true or false? Lack of Consistency does not necessarily constitute evidence against a causal association. )1 mark)

c) Read the following paper and then answer the question below.

Buitrago-Lopez et al. Chocolate consumption and carthometalx>lic disorders systematic review and meta-analysis. BMJ 2011,343 04488

http://www.bmj.com/content/343/bmj.d4488.full

Based on the above meta-analysis address the six main Hill cntena for assessing causation. If it is not possible to assess a criteria based on Me meta-analysis than explain what type ofevidence you would be looking for and the son of study that might be needed to address that cntena. The six main Not criteria are: Expenmental Evidence: Temporality Strength of Associabon.

Dose response: Biological plausibility; and Consistency.

Question 4

Tne following is an extract from the abstract of a published article that investigated neuroblastoma screening in infants. Road the extract and Mon answer the questions below.

Abstract

Background
Nouroblastoma has many characteristics which suggest that preclinical detection might improve outcome. The Quebec Neuroblastoma Screening Project was initiated to determine whether mass screening could reduce mortality in a large cohort of infants. As an early endpoint, we report whether screening could reduce the incidence of poor-prognosis neuroblastoma in children with advanced stage disease over 1 year of age.

Methods

An 476 603 children born in the province of Quebec during the 5-year period of May 1, 1989. to April 30. 1994. were eligible for urinary assay of homovanillic acid and vanillytrnandelic acid at 3 weeks and 6 months of age. Children with a positive screen were referred to one of four paediatric cancer centres in the province for uniform evaluation and treatment if necessary. Standardised incidence ratios (SIRs) were calculated for neuroblastoma in the province and two similar population-based controls. the state of Minnesota and the province of Ontario. during the same penod of time and with similar ascertainment procedures.

Findings

Compliance with screening in Quebec province was 91% at 3 weeks (n=425 816) and 74% at 6 months (n=349 706). Through July 31, 1995, with a follow-up of the birth cohort of 15-75 months. 118 cases of neuroblastoma wore diagnosed. 43 detected preclinically by screening, 20 detected clinically before screening at 3 weeks of age and 55 detected clinically after 3 weeks of age having normal screens (52) or never screened (3). Retrospective analysis of stored samples confirmed that 49 of 52 patients missed by screening had levels of catecholamine metabolites that were too low to be detected at 6 months or earlier. Based on US Surveillance. Epidemiology and End Results data, 54.5 cases of neuroblastoma would have been expected in Quebec province during the study period, for an SIR of 2'17 (95% CI 1.79-2-57, p<0.0001). For the two control groups, 43 and 80 cases of neuroblastoma were detected. respectively, compared with 37.9 and 85.4 expected, overall SIR I.00 (not significant). SIRs for Quebec province by age at diagnosis in yearly intervals show a marked increased incidence under 1 year of age (SIR 2.85. 2.26-3.50), with no reduction in incidence in subsequent years. Limiting analysis to only patients diagnosed over 1 year of age with advanced-stage disease. 22 cases were detected in Quebec province versus 14.4 expected (SIR 1.52, 0-95-2.23) Data in the two control groups show no significant increase or decrease in any-stage disease in children under or over the age of 1 year. except for an increase in early-stage disease in Minnesota children over 1 year. 10 versus 3.8 expected (SIR 2.67. 1-27-4-58).

a) How many cases (in total) of neuroblastoma were detected in Quebec during the study?

b) How many cases were expected to be found in the Quebec province?

c) Were more or less cases found compared to expected in the Quebec province?

d) Did the increased detection in cases in the less than one year age group lead to decreased incidence in the older age groups?

e) What is the most likely explanation for the increased number of neuroblastoma cases found?

Question 5

a) In no more than one paragraph present an example where either an ecological or spatial analysis would be the most appropriate study design. Please give justifications for your answer.