How do I keep a literature review current over a multi-year PhD?

The chapter you write in year one is a chapter about a field that no longer exists by year four. This is normal and it is not a reason to delay writing it. What it does mean is that the chapter should be built so that updating it is cheap, and most review chapters are not.

The expensive version keeps everything in the prose. When a relevant paper appears, you have to find the paragraph it belongs to, work out what the paragraph currently claims, decide whether the new work changes it, and rewrite. Three years of that, and updating feels like starting over, which is why so many people put it off until it is genuinely a rewrite.

The cheap version keeps the evidence outside the prose, in a list, matrix, or map that you add to as you read. New paper, new rows, tagged to the claims it touches, taking about two minutes. The prose is regenerated from the evidence when a claim actually changes. Most new papers will not change a claim, and this structure makes that visible instead of hiding it.

There is a psychological benefit too. Adding a row is a small enough action to do on a bad day, so the review stays current through the periods when writing has stopped. A chapter that is only updated during writing sprints goes stale between them.

This approach has a formal counterpart. Elliott and colleagues describe the living systematic review, a continuously maintained search whose results feed ongoing study identification, quality assessment, data extraction, analysis, and reporting, and they argue that it replaces sporadic and exhausting update efforts with moderate ongoing contributions (Elliott et al., 2014, pp. 2-3). They wrote it for fast-moving health evidence, so if your field publishes slowly, take the continuous intake and lower the frequency. They also caution that reanalysing a synthesis every time new data arrives inflates false-positive findings and produces unstable early effect estimates (Elliott et al., 2014, p. 3), which is a second reason to add rows often and rewrite conclusions rarely.

How often should I rerun my literature searches?

Twice a year for the whole search, and continuously through alerts for your five or six central works. A full rerun every month costs more attention than it returns, and once a year lets a field move without you noticing until the writing-up stage.

The continuous half is the more valuable one. Forward-citation alerts on your key works catch the papers that engage directly with what your argument depends on, which is a much better signal than a keyword alert that returns everything sharing a term with your topic.

Keep the search log from your original strategy and rerun the same strings with a date restriction. That takes minutes rather than hours, produces a manageable result set, and gives you a defensible sentence for the methods section about when coverage ends.

Write down what you searched while you are searching. Pautasso's rules for a literature review include keeping a record of the exact search terms and databases used, adopting a reference-management system from the start, defining exclusion criteria, keeping a list of papers you could not access so you can retrieve them later, and searching earlier reviews as well as primary studies (Pautasso, 2013, p. 1). Pautasso also recommends varying your keywords across several sources, naming Google Scholar, JSTOR, Medline, Scopus, and Web of Science, and tracing citations out of the papers and book chapters you already hold (Pautasso, 2013, p. 2).

How do I update a review chapter six months before submission?

Run the full search once, screen only for papers that would change a claim rather than add to it, and mark those. Most recent work will support what you already wrote and needs a citation at most. Give real revision time only to results that contradict or supersede something in the chapter.

The screening question is narrow on purpose: would the chapter say something different if this paper existed when I wrote it? For the large majority the answer is no, and the honest response is to add the citation to an existing sentence and move on.

Resist the pull to add a "recent developments" section at the end. It reads as an afterthought and it separates new evidence from the claim it bears on. Put each new work where its claim already lives, even when that means touching six paragraphs instead of writing one.

Two more things belong in that final pass. Pautasso notes that publication and database-indexing delays can leave a review out of date by the time it appears, so Pautasso recommends monitoring electronic lists of papers in press and repeating the search at the revision stage, while warning that very recent studies are hard to judge because there is not yet evidence of their significance or influence (Pautasso, 2013, pp. 3-4). The tenth rule is to be up to date without forgetting older studies (Pautasso, 2013, p. 3), so the update pass should protect foundational work rather than crowd it out with whatever appeared last year.

Refresh the methods as well as the reference list. Mayhew, Kabir, and Ansari report that only about 20% of Cochrane reviews were updated every two years, and that of the more than 400 reviews updated in 2014, over 75% had originally been published before January 2011, which left their included studies needing reassessment with newer risk-of-bias methods (Mayhew et al., 2015, pp. 1-2). Their recommendation is to update the appraisal approach itself, not only to add new papers (Mayhew et al., 2015, p. 5). For a chapter written across four years, that means recording which appraisal framework and synthesis decisions you applied at each stage, not just the date of your last search.