Systematic review, 22 studies

Systematic review poster

A systematic review poster has to show how many records were found, how many survived screening, and how much weight the included studies can bear. This A0 landscape example reviews fictional four day week trials in office work: 1,842 records identified, 22 studies included covering 14,309 workers, and output held in 18 of them but only 3 had a concurrent comparison group. It carries an abstract, an introduction, labelled methods, a six box screening flow, one bar chart, a key finding callout and four takeaways, and its flow leaves out one step worth adding before you build yours: the 1,385 records dropped at title screening, the gap between 1,481 screened and 96 full texts.

Create a poster with OneCraftA0 Landscape, printed at 1189 by 841 millimetres

The whole board

The poster at full size, exactly as it prints. Every number, citation and caption on it was written for this example, so the layout is being judged on real content.

Reduced hours with no loss of pay: a systematic review of four day week trials in office based work

A. Lindqvist ¹, R. Mensah ¹, H. Tanaka ²

1 Centre for Work and Organisation, Northgate University · 2 Institute of Labour Economics, Brightwater

ABSTRACT

Four day week trials are widely reported but rarely compared. We searched five databases for trials of a reduced working week with no reduction in pay in office based settings, published between 2015 and 2025, and screened 1,842 records against a written protocol. Twenty two studies covering 14,309 workers met the inclusion criteria. Reported output was maintained or improved in 18 of the 22, but only 6 used an objective output count, and only 3 had a concurrent comparison group. Sickness absence fell in 11 of the 13 studies that measured it, with a pooled reduction of 21%.

INTRODUCTION

The claim that a four day week leaves output unchanged has moved from advocacy into policy, with pilot schemes now run by national governments and by large employers. The primary studies behind that claim are heterogeneous in almost every respect that matters: the length of the trial, whether pay was protected, whether hours were compressed or genuinely reduced, and what counted as output. A review that pools them without recording those differences would produce a number that means very little. We therefore set out to describe the evidence rather than to pool it into a single effect. Our questions were what has actually been measured, how it was measured, whether anything was compared against, and how much of the reported effect rests on workers rating their own productivity.

METHODS

Sources. Five databases plus grey literature from three national pilot programmes, searched to March 2025.

Eligibility. Reduced weekly hours with pay protected, office based work, at least 8 weeks, any comparison design.

Screening. Two reviewers screened independently; disagreements resolved by a third. Kappa was 0.84.

Extraction. Outcome, measurement method, comparison group, trial length and attrition, into a piloted form.

Certainty. GRADE applied per outcome. No meta-analysis of output was attempted given the measurement heterogeneity.

PARTICIPANT FLOW

Records identified

1,842

Duplicates

361

Titles screened

1,481

Full texts assessed

Full texts excluded

Studies included

Figure 1. Included studies by how output was measured. A study can use several methods, so bars sum past 22.

Source: review data extraction, 22 studies, 2026.

RESULTS

Twenty two studies covering 14,309 workers met the criteria. Output was reported as maintained or improved in 18, but the measurement behind that finding is thin: 16 used workers rating their own productivity, 6 used an objective count such as tickets closed or cases handled, and only 3 had a concurrent comparison group rather than a before and after design. Sickness absence was measured in 13 studies and fell in 11, with a pooled reduction of 21% (95% CI 12 to 29). Self reported burnout fell in 14 of 16. Attrition was low across the board, median 4.1%, which is expected when the intervention is a shorter week. Three studies reported an increase in work intensity within the remaining days, measured by self report in all three. Two reported that meeting time was cut rather than task time, which is a different mechanism with different limits on how far it travels. Trial length ranged from 8 weeks to 24 months, with a median of 26 weeks. The two longest trials both reported some regression toward baseline hours by the final quarter, and neither reported whether the reduced week was still formally in place at the end. Nine of the 22 studies were run or funded by an organisation advocating for the policy.

DISCUSSION

The consistent findings in this literature are about how people feel and how often they are absent, not about how much they produce. That is not the same as saying output falls, and no study in the set found a clear drop. It means the confident public claim that output is unaffected rests on 3 studies with a comparison group and 6 with an objective measure. The likely mechanism is fewer meetings, not less task time.

Key finding

Output held or improved in 18 of the 22 studies, but only 3 had a concurrent comparison group.

22 studies, 14,309 workers, output held or rose in 18

Only 6 used an objective output count, 3 had a control

Sickness absence fell in 11 of 13, pooled 21%

Certainty low for output, moderate for wellbeing

REFERENCES

1.Schor JB, et al. The four day week: assessing global trials. World Sociology 2024;12(2):119-148.

2.Delaney H, Casey C. The promise of a four day week: a critical appraisal. Employee Relations 2022;44(1):176-190.

Block by block

What each block on the board is for, in the order a reader walks it.

Title band and authors: The title states the intervention and the scope, reduced hours with no loss of pay in office based work, and names the design as a systematic review so nobody mistakes it for a trial. Three authors sit under it over two affiliations, a university centre for work and organisation and an institute of labour economics.
Abstract and introduction: The abstract gives the five databases, the 2015 to 2025 window, a written protocol, 22 studies and 14,309 workers, output held in 18, only 6 with an objective output count and 3 with a comparison group, and sickness absence falling in 11 of 13. The introduction explains why the studies were described rather than pooled: they differ in trial length, pay protection, whether hours were compressed or cut, and what counted as output.
Methods as a labelled protocol: Five labelled entries. Sources: five databases plus grey literature from three national pilot programmes, searched to March 2025. Eligibility: reduced weekly hours with pay protected, office based work, at least 8 weeks. Screening: two reviewers with a third resolving disagreements, kappa 0.84. Extraction: outcome, measurement method, comparison group, length and attrition into a piloted form. Certainty: GRADE per outcome, with no meta analysis of output. The abstract calls it a written protocol; if yours was registered, add the registry and id here.
The screening flow: Six boxes under the heading Participant flow, which is the fixed heading of this block even when it counts records: 1,842 identified, 361 duplicates, 1,481 titles screened, 96 full texts assessed, 74 full texts excluded, 22 included. The printed subtractions hold: 1,842 less 361 is 1,481, and 96 less 74 is 22. The flow does not show the 1,385 records dropped at title screening (1,481 less 96), and it does not split the 74 full text exclusions by reason.
Figure 1, how output was measured: Four horizontal bars count included studies by measurement: self report 16, manager rating 8, objective output 6 and revenue or billing 3. The bars sum to 33 across 22 studies, and the caption says why: a study can use several methods, so the categories overlap. The abstract and results use the same 16 and 6. The source line names the review data extraction.
Results: The results give the 18 of 22, then the measurement behind it: 16 relied on self rating, 6 used an objective count such as tickets closed, and 3 had a concurrent comparison group. Sickness absence was measured in 13 studies and fell in 11, pooled at 21 per cent with an interval of 12 to 29; burnout fell in 14 of 16; median attrition was 4.1 per cent. It also reports work intensity rising in 3 studies, trial lengths from 8 weeks to 24 months, and 9 studies run or funded by advocates.
Discussion, callout, takeaways and references: The discussion says the consistent findings are about wellbeing and absence, not output, then names the likely mechanism: fewer meetings rather than less task time, which two studies reported. The callout pairs 18 of 22 with the 3 controlled studies. Four takeaways close with the certainty ratings, low for output and moderate for wellbeing, the only place they appear. Two references sit in the footer, one fewer than the block holds; verify every citation detail against the source before printing.
How to adapt this board: Put the counts into the flow in the order the PRISMA 2020 diagram gives them, with title screening exclusions in the aside and full text exclusions split by reason. Keep the overlap note in the chart caption if a study can sit in more than one bar. Add a registry and id to the Sources entry if the protocol was registered, and state each certainty rating in the results next to its outcome.

What makes this board work

The flow subtracts where it prints

Records identified less duplicates gives the 1,481 screened, and full texts less exclusions gives the 22 included. A reviewer checks those two sums first, and both hold. The missing title screening count is the next thing they will ask about, which is why the adapt section puts it back.

The callout carries the caveat

The key finding callout does not stop at output holding in 18 of 22 studies; it adds in the same sentence that only 3 had a concurrent comparison group. A headline that carries its own weakness is harder to misquote than one that leaves the caveat for the results text.

It declines to pool output

The Certainty entry says no meta analysis of output was attempted because the measures were too different, and the introduction explains why a pooled figure would mean very little. Saying what was not done, and why, is stronger than a number nobody can interpret.

Questions people ask

What goes on a systematic review poster?

The question, the databases and search dates, the eligibility criteria, a PRISMA style flow from records to included studies, the synthesis, a certainty rating per outcome and the limits. The protocol registration belongs near the methods. This board has all of these except a registry entry, and its certainty ratings sit in the takeaways rather than beside each result.

Do I need a PRISMA diagram?

For a systematic review, yes. It answers how many records were found, how many were excluded and at which stage, in a form reviewers can check at a glance. Ten of the forty poster designs carry a flow block, so choose one of those before writing, and keep each exclusion stage with its own count.

What if the studies are too different to pool?

Say so in methods, give the reason in the introduction, and chart what the studies are rather than what they found. This board charts how output was measured across the included studies, which explains the spread of results better than a forest plot with no pooled line. Outcomes that were comparable, like sickness absence here, can still be pooled.

Where does the protocol registration go on a review poster?

In the methods, as the first line of the sources or eligibility entry: the registry name and the id, so a reader can check that the eligibility criteria were set before the search. This board describes a written protocol and makes no registration claim; if yours was registered, print the registry and id, or a reader will ask.

Should chart categories on a review poster overlap?

They can, but the caption must say so. If a study can appear under more than one measurement type, the bars will sum past the included count and a reader will think the numbers are wrong. If the categories are exclusive, they must sum to the included count exactly. This chart takes the first route: its caption says a study can use several methods, so the bars sum to 33 across 22 studies.

How do I show certainty of evidence on a poster?

One clause per outcome, next to the result it rates: low for output, moderate for wellbeing, with the main reason for each downgrade. A single overall rating hides the fact that evidence is stronger for some outcomes than others. This layout has no table block, so a list item or a sentence in the results is the practical place.

Build your own in about a minute

The button below opens the generator with this use case already described. Change the wording to match your own, generate, then edit anything you like.

Make one like this

Sources

Written and checked by the OneCraft team. Last checked 29 September 2026.