Dear colleagues,

I am currently considering the Newcastle-Ottawa Scale (NOS) to appraise risk of bias or methodological quality of non-randomized studies (e.g., cohort and case-control) for my systematic review. This tool is based on three broad perspectives: 1) the selection of study participants; 2) the comparability of the study groups; and 3) the identification of either the outcome of interest or exposure for cohort or case-control studies, respectively. However, I noticed that the original tool is grounded on population- or community-based evidence appraisal. Since my subjects are all hospital-based, the following ‘Selection’ subcategories (item number 2, Selection of Non-exposed Cohorts; item number 3, Selection of Controls) for cohort-type and case-control studies were modified:

Newcastle–Ottawa Quality Assessment Scale for Cohort Studies

Category: Selection

Item No.: 2, Selection of Non-Exposed Cohort.

Item Purpose: This item assesses whether the control series used in the study is derived from the same population as the cases and essentially would have been cases had the outcome been present.

Original Criteria:

a. Drawn from the same community as the exposed cohort

b. Drawn from a different source

c. No description of the derivation of the non-exposed cohort

Modified Criteria:

a. Drawn from the sameICU/hospital as the exposed cohort (e.g. exposed and unexposed drawn from the same database or group of patients presenting at same points of care from same hospital over the same or different time frame)

b. Drawn from different source (e.g. exposed and unexposed drawn from the same database or group of patients presenting two different points of care from another hospital over a same or different time frame)

c. No description of the derivation of the non-exposed cohort

Newcastle–Ottawa Quality Assessment Scale for Case–Control Studies

Category: Selection

Item No.: 3, Selection of Controls

Item Purpose: This item assesses whether the control series used in the study is derived from the same population as the cases and essentially would have been cases had the outcome been present.

Original Criteria:

a. Community controls (e.g., same community as cases and would be cases if had outcome)

b. Hospital controls, within same community as cases (i.e. not another city) but derived from a hospitalized population

c. No description

Modified criteria:

a. ICU-based controls (e.g. same hospital/ICU as cases and would be cases if had outcome) 


b. Hospital controls, within same or similar ICU-type settings as cases (i.e. not another different ICU type) but derived from another hospital

c. No description

I would like to seek opinions from a randomly select group of international experts with extensive experience in using NOS to validate whether the modified items are appropriate for an hospital-based review. Hope to hear from you.

Thank you,

John

Note: bold and Italicized words/statements are modified items