A literature review starts with a focused research question and a documented search, not with a pile of articles. The rest follows in order: you screen what the search finds against clear criteria, record each study's design and findings in an evidence matrix, group the findings into themes, and only then write. Most guides describe these steps in general terms. This literature review example works through all of them on a real topic, with real searches, real study results and the decisions made at each stage.
The topic is a common one in nursing and health sciences: whether bed and chair alarms prevent falls in hospital. The searches were run in PubMed on October 12, 2026, so the numbers you get today may differ slightly. Use the process for any subject, from education to psychology to business.
"Falls in hospitals" is a topic, not a question. Researchers have studied exercise, medication review, staff education, low beds, hourly rounding and many other strategies, so a review of "falls" would never end. The first job is to narrow it to one question that a body of studies can answer.
For clinical topics, the PICOT format does this well. Our guides on how to formulate PICOT questions and weak vs. strong PICOT examples explain the method in detail.
| Element | This review |
|---|---|
| Population | Older adults (65 and over) in acute or subacute hospital wards |
| Intervention | Bed or chair alarms and pressure sensors that alert staff when a patient gets up |
| Comparison | Usual care without alarms, or a different alarm type |
| Outcome | Falls per 1,000 bed-days, number of fallers and fall injuries |
| Time | The hospital stay |
Research question: In older adults admitted to hospital, do bed and chair alarms reduce falls compared with usual care?
Split the question into concepts, list synonyms for each, and find the database's controlled vocabulary. In PubMed that vocabulary is MeSH. The comparison and time elements usually stay out of the search because they are judged when you read each study.
| Concept | Keywords | Subject heading |
|---|---|---|
| Falls | fall, falls, falling | Accidental Falls [MeSH] |
| Alarms | bed alarm, bed exit, bed sensor, pressure sensor, chair alarm, chair sensor | None fits closely, so keywords carry this concept |
| Hospital setting | hospital, inpatient, ward | Hospitals [MeSH] |
Synonyms within a concept are joined with OR, and concepts are joined with AND. The truncation symbol * picks up word endings, and quotation marks keep phrases together. The final PubMed string was:
("Accidental Falls"[Mesh] OR fall*[tiab]) AND ("bed alarm*"[tiab] OR "bed exit"[tiab] OR "bed sensor*"[tiab] OR "pressure sensor*"[tiab] OR "chair alarm*"[tiab] OR "chair sensor*"[tiab]) AND (Hospitals[Mesh] OR hospital*[tiab] OR inpatient*[tiab] OR ward*[tiab])
The tag [tiab] limits a keyword to titles and abstracts, which cuts down on articles that mention a term only in passing. For a full review you would run an adapted version of this search in CINAHL and the Cochrane Library as well, because each database indexes different journals.
Record the date, database, exact search string and number of results at each stage. Your instructor, or anyone reading your review, should be able to repeat the search. Here is how the numbers changed as the search was refined:
| Search stage | PubMed results |
|---|---|
| Plain-language search: falls AND bed alarms | 92 |
| Falls concept AND alarms concept, with MeSH and synonyms | 224 |
| Plus the hospital setting concept | 64 |
| Plus English language | 64 |
| Plus trial and review designs (randomized trials, controlled trials, systematic reviews and meta-analyses) | 22 |
Notice that the structured search found more records than the plain-language one. Synonyms such as "bed exit" and "pressure sensor" pick up studies that never use the phrase "bed alarm." Adding the setting then removed studies done in nursing homes and private homes.
Write your inclusion and exclusion criteria before you read the results. That stops you from choosing only the studies that agree with what you expected.
Screening the 22 trial and review records by title and abstract produced several clear exclusions. They included a 2011 protocol for a trial whose results were published in 2016, a 2005 paper on heart failure that matched the search by accident, and a review about reducing physical restraints rather than falls. Eight sources were kept, listed in the matrix below.
For a class assignment of 8 to 15 sources, this level of screening is usually enough. A formal systematic review would screen all 64 records, search two or three more databases, check reference lists, and report the numbers in a PRISMA flow diagram.
An evidence matrix is a table with one row per source and a column for each thing you need to compare. It turns reading notes into something you can synthesize. These are the eight sources from this search, with figures taken from each study's published abstract:
| Study | Design and setting | Sample | Intervention vs. comparison | Key finding |
|---|---|---|---|---|
| Shorr et al. (2012) | Cluster randomized trial, 16 units, one US hospital, 18 months | 27,672 inpatients | Education and support to increase bed alarm use vs. alarms available but not promoted | Alarm use rose sharply (64.41 vs. 1.79 days per 1,000 patient-days), but fall rates did not change (risk ratio 1.09, 95% CI 0.85 to 1.53) |
| Sahota et al. (2014), REFINE | Randomized controlled trial, acute medical wards, UK teaching hospital | 1,839 older patients | Bed and chair pressure sensors linked to pagers vs. standard care | 8.71 vs. 9.84 bedside falls per 1,000 bed-days, not significant (IRR 0.90, 95% CI 0.66 to 1.22); not cost-effective |
| Barker et al. (2016), 6-PACK | Cluster randomized trial, 24 acute wards in 6 Australian hospitals, 12 months | 31,411 patients | Nurse-led program of six options, including bed/chair alarms, vs. usual care | Practice changed, but falls (IRR 1.04, 95% CI 0.78 to 1.37) and fall injuries were similar |
| Cameron et al. (2018) | Cochrane systematic review of 95 trials; 24 in hospitals | 138,164 participants overall | Bed and chair sensor alarms in hospitals pooled from 2 trials (28,649 participants) | Effect uncertain (rate ratio 0.60, 95% CI 0.27 to 1.34), very low-quality evidence |
| Morris et al. (2022) | Systematic review and meta-analysis of hospital falls interventions | 43 studies, 23 meta-analyzed | Education, environment, devices, systems and others | Only education significantly reduced falls (rate ratio 0.70); bed alarms, chair alarms and wearable sensors were not associated with fewer falls |
| Hempel et al. (2013) | Systematic review of US acute care hospital programs | 59 studies | Mostly multicomponent programs, many with bed-exit alarms | Pooled IRR 0.77 (95% CI 0.52 to 1.12), not significant, with very high heterogeneity; reporting of implementation and adherence was poor |
| Eichenbrenner et al. (2026) | Non-randomized controlled study, delirium unit, Switzerland | 119 patients, 1,799 patient-days | Contactless 3D-radar bed-exit monitoring vs. contact floor mat | Bed-exit falls 11.5 vs. 22.6 per 1,000 patient-days, favoring radar; fewer alarms and less nurse time |
| Considine et al. (2023) | Qualitative interviews, 32-bed geriatric ward, Australia | 12 nurses | Nurses' experience of bed and chair alarms | Experiences mostly negative: noisy, imperfect technology and a tension between limited safety benefit and the risks of overuse |
Add columns for anything your assignment asks you to judge, such as level of evidence, risk of bias or limitations. Our guide on the limitations of a study shows how to describe them.
Synthesis means reading down the columns of your matrix, not across the rows. Instead of asking what each study found, ask where the studies agree, where they disagree, and why. Four themes come out of this matrix.
The gaps become clear from these themes: there are few randomized trials of newer sensor technology, studies are mostly single-site, and patients with delirium or dementia, who may benefit most, are rarely studied separately.
Organize the review by theme, not by study. Each body section should draw on several sources.
Here is how the first body section might read. Notice that each sentence makes a point and that sources support the point together, instead of each source getting its own paragraph.
Bed and chair alarms are widely used in hospitals, yet randomized evidence does not show that they prevent falls when used on their own. In a cluster randomized trial covering more than 27,000 inpatients, an intervention that greatly increased bed alarm use produced no change in fall rates (Shorr et al., 2012). A UK trial that randomized 1,839 older patients to bed and chair pressure sensors found a similarly non-significant difference and concluded that the sensors were not cost-effective (Sahota et al., 2014). Systematic reviews reflect the same uncertainty: the Cochrane review rated the evidence on hospital sensor alarms as very low quality (Cameron et al., 2018), and a later meta-analysis found that alarms and wearable sensors were not associated with fewer falls, whereas patient and staff education was (Morris et al., 2022). Taken together, these studies suggest that alerting staff when a patient gets up does not prevent the fall unless staff can respond in time, a point echoed by nurses who describe alarms as frequent, noisy and often false (Considine et al., 2023).
| Mistake | Why it weakens the review | Fix |
|---|---|---|
| One paragraph per study | It reads like an annotated bibliography, with no synthesis | Organize by theme and cite several studies per point |
| Not recording the search | Readers can't judge how complete it is | Report databases, dates, strings and result counts |
| Treating all studies as equal | A single-site pilot carries less weight than a large trial | Mention design and sample size when you compare findings |
| Ignoring studies that disagree | It biases the conclusion | Explain differences in setting, population or design |
| Ending with a summary only | It misses the purpose of the review | State the gaps and what they mean for practice or research |
If you're also asked to appraise individual studies, our guide on how to write an article critique covers what to look for in each one.
Literature reviews take time because every step depends on the one before it. If your search is returning too many or too few results, or your draft still reads like a list of summaries, it helps to have someone check your strategy and structure. Our nursing homework help and research paper help connect you with writers who can review your question, search and outline. Use any outside help in line with your school's academic integrity policy.
It depends on the assignment. Course literature reviews often use 8 to 20 sources, while theses and systematic reviews use many more. Your instructor's rubric is the best guide, but quality and relevance matter more than the count.
It's a table with one row per source and columns for details such as design, sample, intervention, findings and limitations. It helps you compare studies and find themes before you write.
Most reviews are organized by theme, because that shows how studies relate to each other. A chronological structure can work for showing how a field developed, but it tends to produce one paragraph per study.
Many instructors expect a short one describing the databases, search terms, dates and inclusion criteria. Systematic reviews require a detailed methods section and a PRISMA flow diagram.
An annotated bibliography summarizes and evaluates each source separately. A literature review synthesizes sources into themes and draws conclusions about what the evidence shows as a whole.
Databases add new records every day and sometimes update their indexing, so counts change over time. That is why you should record the date of each search.
Get affordable and top-notch help for your essays and homework services from our expert tutors. Ace your homework, boost your grades, and shine in online classes—all with just a click away!
Fast, secure, and handled by vetted experts.
Is AI homework help as good as a human tutor? We compare both across 10 learning tasks using published research, with clear verdicts and tips for using AI well.
Read More
Learn how to write a PICOT question in six steps, with templates for every question type, a worked nursing example, search tips and common mistakes to avoid.
Read More
What makes a good research question? Learn the key criteria, the main question types and how to write your own, with weak vs. strong examples. Start drafting.
Read More