Screening is the stage where a retrieval becomes a study set, and it is graded on process rather than on outcome. The written work reports how records moved from thousands to a handful: the two-stage sift at title and abstract and then at full text, the criteria applied at each stage, how many people screened and how disagreements were settled, the agreement achieved, and a reason recorded against every full-text exclusion. Done properly this section is dull, numerate and completely convincing. Your section may print this as NR 714 or NR714; it is the same course. Chamberlain publishes no syllabi outside Canvas. The placement here is our teaching judgment from the course's catalog arc; your section's rubric decides what your week actually asks.
What NR-714 Week 3 asks for
Two reviewers sit down with fourteen hundred titles on delirium screening at the point of transfer between hospitals and skilled nursing facilities. They agree on almost everything and disagree on forty-one records, and the disagreements cluster in one place: studies where screening happened at admission to the facility rather than at discharge from the hospital. That cluster is not noise. It is a criterion that was written ambiguously in the protocol and is now being interpreted two ways, and the correct response is to resolve the definition, record the amendment and rescreen the affected records rather than to split the difference case by case.
Screening has two stages with different standards of evidence. At title and abstract you are excluding only what is clearly ineligible, and the correct instinct is permissive: when the abstract is ambiguous, the record moves forward. At full text you are applying every criterion strictly, and here every exclusion needs a recorded reason drawn from a fixed list, because those reasons are reported in the flow account and readers use them to judge whether your criteria were applied consistently.
Agreement is worth reporting for two reasons. Practically, disagreement rates tell you whether criteria are ambiguous while there is still time to fix them. Methodologically, a reported agreement statistic converts screening from an assertion into a measured process. Percentage agreement is easy to compute and overstates performance because it takes no account of chance; a chance-corrected statistic such as kappa is the standard reported alongside it. Whatever you report, report the raw counts too, since a kappa computed on a set where nearly everything was excluded behaves strangely and the counts let a reader see that.
Two record-keeping practices decide how painful the later stages become. First, keep a single deduplicated record list with a decision column and a reason column, rather than filtering the same list repeatedly. Second, log the full-text records you could not obtain and how you tried, because unretrievable studies belong in the flow account rather than in silence. Expect a written screening report with a flow account, an exclusion reason table, and often a discussion post about a screening decision. Posts do not reopen once submitted in Canvas, so state the criterion rather than the impression.
The NR-714 Week 3 method, step by step
Six moves from a pile of records to a defensible study set.
-
Calibration on a pilot set before full screening
Screen the same fifty records independently, compare, and repair the criteria that produced disagreement. An hour spent here saves days later and produces the sentence about calibration that your methods section needs.
-
Application of the permissive rule at title and abstract
Exclude only what is clearly ineligible and carry ambiguity forward. Records wrongly excluded at this stage are invisible afterwards, while records wrongly retained cost only a few minutes at full text.
-
Retrieval of full texts with a documented trail
Track requests, interlibrary loans and failures. Report the number you could not obtain and what you attempted, because unretrieved records are a limitation that reviewers expect to see acknowledged.
-
Strict assessment at full text with a single recorded reason
Apply criteria in a fixed order and record the first one that fails: wrong population, wrong intervention, wrong comparator, wrong outcome, wrong design, duplicate report. One reason per record keeps the exclusion table interpretable.
-
Resolution of disagreements by a stated rule
Discussion first, third party if unresolved, and a note if a criterion had to be clarified. Where clarification changes a rule, rescreen the affected records rather than leaving inconsistent decisions in place.
-
Reconciliation of every count in the flow account
Records identified, duplicates removed, screened, excluded, sought, unretrieved, assessed, excluded with reasons, included. The numbers must add up exactly, and grouped reports of one study must be counted as one study with its companion papers linked.
A layout and word budget for a screening report
Our frame for the selection deliverable, sized for roughly 1,300 to 1,700 words plus the flow account and exclusion table. It is our own outline rather than anything the university issues, and your week's rubric outranks it wherever the two disagree.
| Section | What belongs in it | Word target |
|---|---|---|
| Screening design | The stages, the number of screeners at each, the software or system used, and the calibration performed. | 200 to 250 |
| Criteria as applied | The operational form of each criterion at the point of use, including any clarification made during screening. | 240 to 300 |
| Title and abstract stage | Records screened, the permissive rule applied, disagreement count and how it was resolved. | 200 to 260 |
| Full-text stage | Records sought, records unretrieved with the attempts made, records assessed and the strict application of criteria. | 220 to 280 |
| Agreement and exclusions | The agreement statistic with its raw counts, and a table of exclusion reasons with a frequency for each. | 240 to 300 |
| Included set described | How many studies, how many reports, their designs and settings in one paragraph, and what the set implies for synthesis. | 220 to 280 |
Evidence craft for selection reporting
Count studies, not papers. A single trial reported in three publications is one study with three reports, and treating each paper as an independent study inflates the evidence base and can double-count participants in any pooled result.
Give every full-text exclusion exactly one reason. Multiple reasons per record make the table impossible to total. Apply criteria in a fixed order and record the first failure, and state that ordering in the text so the table can be interpreted.
Report agreement with the raw numbers behind it. Give the number of records screened, the number of disagreements and the statistic together. Where screening was done alone, say so and describe what you did instead, such as a second pass on a sample after an interval.
Record clarifications as protocol amendments. When a criterion turns out to be ambiguous mid-screening, write down the clarification, the date and the reason. This is expected practice; silent reinterpretation is what damages a review.
Keep near-miss studies visible. Studies excluded at full text for a narrow reason often belong in the discussion, because a reader familiar with the field will look for them. Naming a few in the text pre-empts the question of whether you saw them.
Five mistakes that cost points in this week's territory
- Numbers that do not reconcile. If the flow account does not add up, every subsequent section is read with suspicion, and the error is trivially easy for a grader to check.
- Strict screening at title stage. Excluding aggressively on abstracts loses eligible studies permanently and invisibly, which is the one screening error that cannot be corrected later.
- Exclusion reasons written as prose. A paragraph explaining why various studies were dropped cannot be counted or checked. The reasons belong in a table with frequencies.
- Companion papers counted as separate studies. This inflates the evidence base and risks double-counting the same participants in synthesis.
- Agreement claimed without a statistic or a count. Saying the reviewers agreed on most records is an impression, and this stage is graded on measurement.
Before you submit
- Calibration on a pilot set is described
- Every stage reports its screener count and disagreement rule
- Unretrieved full texts are reported with the attempts made
- Each full-text exclusion carries exactly one reason from a fixed list
- Agreement is reported with the raw counts behind it
- Studies and reports are counted separately and the flow account reconciles
- Every reference appears in the text and every in-text citation appears in the list
Writing up NR-714 screening this week?
Send the rubric and your record counts out of Canvas. A premium original draft comes back in 24 to 48 hours with a reconciled flow account, a countable exclusion table and agreement reported properly, and revisions run until the grade lands.