Critical appraisal of survival paper I must make a critical appraisal of a cancer survival paper. I hope someone here can give some hints. The paper investigates person-level markers of socio-economic status as exposure and looks for association of these with survival and also comorbidities, disease/prognostic factors and treatment using Cox proportional hazards for survival and logistic regression for mortality.  
I think the paper is a good one and I can't find the criticisms. I am not sure if Cox proportional hazards is the best model for use; maybe a full paramteric model of relative survival is better, but I don't known. 
Thank you for a hint if you can. 
http://www.ncbi.nlm.nih.gov/pubmed/22315055
 A: I think they must have miswritten the cumulative hazard assumption - it should be the proportional hazard assumption. You can probably elaborate some about testing assumptions: will a test for an assumption depend on sample size?
Another hint I can give is the investigation of non-linearity by $age^2$ - what happens at the ends of the parabola and what alternatives do you have for testing the linearity assumptions?
It is also good to question if they should have used an adjustment for multiple testing - see Bonferroni-Holm
I don't think a parametric method would have been preferred.
Regarding general article appraisal of articles I always look at:
- Selection bias - who's in the study
- Information bias - what goes into the variables studied
- Outcome
This should give enough for 1500 words
A: Max makes some excellent suggestions. To supplement his suggestions, I suggest that you read the STROBE and REMARK guidelines. The former gives general advice on the conduct and dissemination of observational studies. The latter specifically deals with studies of molecular prognostic markers, but is still widely applicable to the paper that you have to review. According to Table 1 of the REMARK guidelines, which I directly quote below, the statistical reviewer should assess the following:
METHODS
Study design


*

*State the method of case selection, including whether prospective or retrospective and whether stratification or matching (e.g., by stage of disease or age) was used. Specify the time period from which cases were taken, the end of the follow-up period, and the median follow-up time.

*Precisely define all clinical endpoints examined.

*List all candidate variables initially examined or considered for inclusion in models.
Give rationale for sample size; if the study was designed to detect a specified effect size, give the target power and effect size.


Statistical analysis methods


*

*Specify all statistical methods, including details of any variable selection procedures and other model-building issues, how model assumptions were verified, and how missing data were handled.

*Clarify how marker values were handled in the analyses; if relevant, describe methods used for cutpoint determination.


RESULTS
Data


*

*Describe the flow of patients through the study, including the number of patients included in each stage of the analysis (a diagram may be helpful) and reasons for dropout. Specifically, both overall and for each subgroup extensively examined report the numbers of patients and the number of events.

*Report distributions of basic demographic characteristics (at least age and sex), standard (disease-specific) prognostic variables, and tumor marker, including numbers of missing values.


Analysis and presentation


*

*Show the relation of the marker to standard prognostic variables.

*Present univariate analyses showing the relation between the marker and outcome, with the estimated effect (e.g., hazard ratio and survival probability). Preferably provide similar analyses for all other variables being analyzed. For the effect of a tumor marker on a time-to-event outcome, a Kaplan–Meier plot is recommended.

*For key multivariable analyses, report estimated effects (e.g., hazard ratio) with confidence intervals for the marker and, at least for the final model, all other variables in the model.

*Among reported results, provide estimated effects with confidence intervals from an analysis in which the marker and standard prognostic variables are included, regardless of their statistical significance.

*If done, report results of further investigations, such as checking assumptions, sensitivity analyses, and internal validation.

A: It is hard to improve on the great advice already given but I want to give it a try.  I don't have specific points to raise or articles to suggest but if you need more help on the pros and cons of the Cox model and alternatives I have two gret books to suggest (sorry if you are already very familiar with them).
1. Applied Survival Analysis 2e by Hosmer, Lemeshow and May (2008) Wiley.  I just attended a short course by Susanne May based on this book.  It has really good coverage of survival methods, parametric, nonparametric and the semiparametric Cox model and includes some fairly new material on variable selection in Cox regression.  It is very applications oriented and includes real examples with heavy use of the data from the Worchester Heart Attack Study.  it covers extensions of the Cox model and handling missing data.
2. Modeling Survival Data: Extending the Cox Model by Therneau and Grambsch (2000) Springer-Verlag.  This is all about the Cox model and its extensions but is a little less current.  A second edition is about to come out I think (if it hasn't already).
