Methodological quality criteria in observational studies and clinical trials5.6.
| Observational Studies | Rationale |
|---|---|
| A. Completeness of patient descriptions* | |
| 1. Were American College of Rheumatology classification criteria for SLE and lupus nephritis used for enrollment? | Description of the American College of Rheumatology classification criteria for SLE and lupus nephritis allows standardization of patient inclusion and appraisal of the accuracy of diagnosis. |
| 2. Were measures of renal function at study entry provided? | Description of the patient characteristics is important for determining representativeness and possible selection bias. |
| 3. Were data on treatment provided? | Description of medication use is important for determining its potential effect on the outcomes. |
| B. Representativeness* | |
| 4. Was an inception cohort studied? | In inception cohorts, patients are identified at an early and uniform point in the course of their disease, minimizing bias that may occur by omitting patients who meet the study outcome shortly after diagnosis. |
| 5. Was the cohort community-based? | Recruiting community-based samples of patients can help to achieve greater representativeness compared to cohorts based at referral centers. |
| C. Adequacy of followup* | |
| 6. Were losses to followup reported? | Loss to followup can affect the study validity because patients lost to followup may have different outcomes from those who complete the study. |
| 7. Were losses to followup < 20%? | Loss to followup of > 20% of patients can seriously affect the validity of results because variant outcomes in this subset are often large enough to affect the overall study estimate. |
| 8. Were time-to-event data provided as Kaplan-Meier plots? | Kaplan-Meier plots provide a complete history of events in the study, and it is the standard method to present time-to-event data. |
| Clinical Trials** | Rationale |
|---|---|
| 1. Sequence generation (methods of random assignment to intervention groups) | Random assignment of patients can ensure equivalent-group comparison, eliminating selection and confounding biases. |
| 2. Allocation concealment (methods of concealing the allocation sequence from those enrolling patients) | Allocation concealment prevents selection bias. |
| 3. Blinding of participants, personnel and outcome assessors (a practice of keeping patients and personnel unaware of which intervention is administered to which participant. In cases of no blinding, the likelihood that open-label treatment could influence the outcomes should be low) | Blinding is important because knowledge of which group the patient has been assigned to may affect the outcome. |
| 4. Incomplete outcome data (missing data due to attrition or due to exclusion from the analysis) | Missing data may introduce bias if the number and characteristics of people lost to followup differ between the groups. |
| 5. Selective outcome reporting (information about whether outcome reporting is sufficiently complete and transparent) | Reporting of some outcomes but not others introduces bias. |
↵* Adapted from Wells GA, et al. [Internet.]: www.ohri.ca/programs/clinical_epidemiology/oxford.asp.
↵** Adapted from Higgins JP, et al. BMJ 2011;343:d5928. SLE: systemic lupus erythematosus.