Sage Journals: Discover world-class research

Abstract

Objective

To synthesize the self-reported prevalence of scientific misconduct in health-related research through a systematic review and meta-analysis, including a broader range of unethical practices than previous investigations.

Methods

A systematic search was conducted across six electronic databases and three sources of gray literature. Observational studies reporting the prevalence of scientific misconduct among researchers, students, or health professionals were included. The methodological quality was assessed using the Joanna Briggs Institute checklist, and a random-effects meta-analysis was performed using the Freeman-Tukey double arcsine transformation.

Results

Twenty-nine studies were included, totaling 17,919 analyzed responses. The most prevalent form of misconduct was honorary authorship (32%; 95% CI: 21%–44%), followed by plagiarism (28%; 95% CI: 12%–47%), data falsification (12%; 95% CI: 6%–19%), data fabrication (11%; 95% CI: 6%–19%), and methodological manipulation (4%; 95% CI: 2%–7%). High heterogeneity was observed for most outcomes (I² ≥ 95%), except for methodological manipulation (I² = 62%). No evidence of publication bias was found (p > 0.05).

Conclusion

Scientific misconduct remains a widespread issue in health research. The findings underscore the need for institutional policies that promote research integrity, transparent authorship criteria, and ethical training. Despite the potential for underreporting due to self-reported data, the results provide an important overview of the extent and types of misconduct present in the current scientific practice.

Keywords

scientific misconduct research ethics scientific integrity scientific publications scientific fraud

Introduction

Scientific misconduct refers to the manipulation of collected and analyzed data, which may compromise the results of the research, reducing or nullifying the veracity of the information presented.¹ This practice compromises the reliability of the science and researchers involved, as well as posing risks to the population, especially in areas such as health, in which decisions are largely evidence-based.²

Data manipulation can occur intentionally or accidentally, resulting from methodological failures or data analysis. Among the forms of scientific misconduct are the falsification of data, the manipulation of methods and results, and plagiarism. Plagiarism comprises the unreferenced use of parts or all of third-party studies, while self-plagiarism refers to the reuse of own studies previously published without proper citation.³

Another factor that may compromise data integrity is the conflict of interest, which occurs when personal, professional, or institutional interests influence the researcher’s judgment. In such cases, there may be intentional distortion of the results in order to meet certain expectations or objectives. Given the relevance of the theme, different initiatives have been proposed by scientific and editorial institutions to curb these practices and preserve the integrity of research.⁴

A systematic review with meta-analysis investigated the prevalence of scientific misconduct, both self-reported and non-self-reported, in the biomedical area. The analysis focused on practices such as falsification, data fabrication and plagiarism, revealing high rates, with a significantly higher prevalence among non-self-declared cases.⁵ However, the search strategy adopted was not comprehensive, limited to the consultation of a single indexed database, which may have compromised the representativeness of the results. In addition, scientific misconduct remains a highly relevant issue, as it undermines the reliability of research, weakens public trust in science, and can negatively influence clinical decisions and public policies. Although widely discussed, its true extent is difficult to determine because many cases are not detected or formally reported. Growing academic pressure to publish and evaluation systems focused on productivity rather than research integrity may contribute to the persistence of such practices. Therefore, expanding the current evidence with broader and more rigorous approaches is essential to better understand the phenomenon and support actions that promote research integrity.^6,7

Therefore, this systematic review aimed to synthesize the evidence on the prevalence of scientific misconduct in health research and to expand the scope of analysis by including a broader range of inadequate practices in scientific work. In addition, it sought to answer the following research question: What is the self-reported prevalence of scientific misconduct among researchers, students and health professionals involved in health-related research?

Methods

The protocol of this systematic review was elaborated according to the guidelines of the Preferred Report Items for Systematic Reviews and Meta-Analysis (PRISMA).⁸ The registration was carried out in the International Prospective Register of Systematic Reviews (PROSPERO) database under number CRD42021245251.

Eligibility criteria

The eligibility criteria were defined according to the PEOS framework. We included observational cross-sectional studies that investigated the prevalence of scientific misconduct among researchers, professionals, and undergraduate students in the health field, regardless of their specific training. Eligible studies had to involve participants actively engaged in health research and report the self-reported prevalence or frequency of practices such as data fabrication, data falsification, plagiarism, and honorary authorship, whether committed or witnessed by the respondents. No restrictions were applied regarding gender, ethnicity, or age group. Studies were excluded if they did not involve health researchers, did not provide prevalence data, focused exclusively on knowledge, perceptions, or understanding of scientific misconduct without reporting the prevalence of committed or witnessed practices, or belonged to other publication types, such as reviews, letters, books, conference abstracts, case reports, case series, opinion articles, or technical reports.

Sources of information and research strategy

For the identification of the studies, structured searches were performed using controlled descriptors and free terms in the following databases: Latin American and Caribbean Literature in Health Sciences (LILACS), PubMed/MEDLINE, Scopus, LIVIVO, Embase and Web of Science. Gray literature was explored through sources such as Google Scholar, OpenGrey and ProQuest Dissertation & Thesis. Additionally, manual searches were performed in the reference lists of the included studies (Figure 1), with the help of the Citation Chaser software. The EndNote X20 reference manager (Thomson Reuters, Philadelphia, PA) was used to import references and eliminate duplicates. Database searches were conducted in December 2023 and updated in March 2025. The complete search strategies for all databases are detailed in Appendix 1.

Figure 1.

Flowchart of literature search and selection criteria. *Consider, if feasible to do so, reporting the number of records identified from each database or register searched (rather than the total number across all databases/registers). **If automation tools were used, indicate how many records were excluded by a human and how many were excluded by automation tools. Source: Page MJ, et al. BMJ 2021;372:n71. doi: 10.1136/bmj.n71.

In addition, the term ‘epidemiology’ was included to increase search sensitivity and identify observational surveys reporting prevalence or frequency of scientific misconduct, even when such studies were not explicitly labeled as surveys or questionnaires.

Selection process

The selection of the studies occurred in two phases. In phase 1, two independent reviewers blindly examined the titles and abstracts of all references identified in the electronic databases, based on the previously defined inclusion criteria. Studies that did not meet the criteria were excluded. In phase 2, the same reviewers evaluated, also independently and blindly, the full texts of potentially eligible articles. Both reviewers analyzed the reference lists of the included studies manually. To enable independent screening in both stages, the Rayyan Intelligent Systematic Review (https://www.rayyan.ai) website was used. Although the Rayyan platform includes optional artificial intelligence tools to suggest study inclusion or exclusion, these functionalities were not used. The software was employed exclusively to blind decisions between reviewers and to organize the screening process; all eligibility judgments were performed manually by the reviewers.

In all evaluations, reviewers remained blind to each other’s decisions, and any disagreements were resolved by consensus with the participation of a third reviewer. To ensure adequate calibration among the evaluators, the Kappa coefficient of agreement was previously calculated, and the systematic reading was started only after obtaining a value over 0.7, indicating satisfactory agreement.

Data collection process

Two reviewers independently extracted data from the manuscripts using a pre-defined data extraction form. Any discrepancies were resolved by discussion or, where necessary, by consulting a third reviewer. The following information was collected: authors, year of publication, country of origin, title, abstract, typical characteristics of the sample (including size and origin), qualifications of the interviewees, most prevalent behaviors observed, and the criteria used for their evaluation.

Data items

Data concerning the frequency (absolute or relative) of scientific misconduct practices committed or witnessed were extracted from the included studies, as well as the corresponding sample size of each study and of each outcome evaluated, according to the availability of the information in the respective articles.

Assessment of the methodological quality

The methodological quality of the included studies was evaluated by two independent reviewers using the Joanna Briggs Institute Critical Appraisal Checklist for cross-sectional studies.⁹ The methodological quality was classified into four categories: low, moderate, high and not applicable. The classification was determined according to the proportion of “yes” responses in the evaluated domains: high quality when ≥70% of responses were “yes,” moderate between 50% and 69%, and low when ≤49%. After the evaluation, the results were compared between the reviewers, and potential differences were resolved by consensus, with the participation of a third reviewer, when necessary. The graphical visualizations were generated with the help of the Robvis website (https://mcguinlu.shinyapps.io/robvis/).

Effect measures and synthesis method

A random-effects meta-analysis was conducted to estimate the proportion of scientific misconduct practices in health research, weighted by the inverse of the variance of each included study. The choice of the random-effects model was based on the expectation of methodological and population variability among studies, implying that the true effect may differ across them. This rationale is consistent with the recommendations of the JBI Manual for Evidence Synthesis,¹⁰ which advocates for the use of this model when clinical, methodological or statistical heterogeneity is present and when more generalizable estimates are desired. Between-study variance (Tau²) was estimated using the DerSimonian and Laird method, and heterogeneity was assessed using Higgins’s inconsistency index (I²), which quantifies the proportion of variability across studies attributable to real differences rather than chance. To ensure the approximation of normality in the distribution of proportions—particularly in studies with values close to 0 or 1—the Freeman–Tukey double arcsine transformation was applied, as recommended for meta-analyses of proportions with substantial variability. The 95% confidence intervals were calculated using the exact Clopper–Pearson method, which provides greater precision for binary estimates, especially in small samples or when events are rare.

Potential sources of heterogeneity were explored through subgroup analyses and meta-regression using random-effects models to assess whether variability in the pooled prevalence estimates could be explained by the available study-level covariates. Subgroup analyses were performed when the relevant categorical variables were available across studies, while meta-regression was applied to continuous covariates when sufficient data were reported. A significance level of 5% was adopted. All statistical analyses were performed using RStudio software (Boston, USA).

Reporting bias

Publication bias was evaluated through visual inspection of the funnel chart and the application of the Egger test, considering p values below 0.05, indicating asymmetry.

Results

Selection of studies

Five thousand one hundred and thirty-six articles were initially identified in the databases. After the removal of duplicates, 3791 references were left for the screening of titles and abstracts. Of these, 67 were selected for full reading (phase 2), resulting in the exclusion of 38 studies, as detailed in Appendix 2. None of the studies from the gray literature and the list of references met the inclusion criteria. (Figure 1).

Characteristics of the studies

The included studies were published in English and conducted in several countries, including South Africa, Germany, Canada, China, Denmark, Egypt, the United States, Netherlands, India, Iran, Iraq, Israel, Jordan, Nigeria, Pakistan, Kenya, the United Kingdom, Serbia, Sweden and Switzerland. The outcomes analyzed referred to different forms of scientific dishonesty, such as data fabrication and falsification, methodological manipulation, plagiarism and honorary authorship.

The data collection was performed anonymously, through questionnaires applied to health professionals, students and researchers. In the studies that reported the gender of the participants, there was a predominance of females. The age group of the respondents varied between 19 and 77 years.

The included studies showed substantial variability in terms of professional field (students, researchers, and health professionals), institutional context, and survey focus, as reported by the original studies.

Methodological quality of the studies

Only one study was classified as having low methodological quality, while another was rated as moderate; the remaining studies were considered to have high methodological quality (Figure 2). The main methodological weaknesses identified included: absence of detailed statistical techniques used; lack of specific information on the validity and reliability of exposure variables measures; unclear inclusion criteria; lack of description of limitations of the studies; lack of identification of potential confounding factors; and lack of strategies for their control.

Figure 2.

Forest plot of the prevalence of research misconduct and the methodological quality of each included study.

Individual results of the studies

Most participants in the included studies had higher-level degrees, except for a study that mentioned the participation of individuals without an academic background.¹¹ Two articles did not provide information about the level of education of respondents.^12,13 Among the most frequently reported behaviors in the studies were data fabrication, falsification and plagiarism. In total, 19 articles described at least one of these practices of scientific misconduct.^{11,12,14–30} Plagiarism was pointed out as the most common infraction in seven studies, with expressive percentages compared to the other behaviors evaluated.^{12,16,17,20,21,24,25} In five other studies, data fabrication was identified as the most prevalent practice,^{11,14,19,23,28} often associated with information falsification.^{12,16,21,23,25,28}

Questions related to authorship were also highlighted, as the main focus of nine studies.^{2,12,13,15,20,27,28,31–33} Among the most frequently reported situations were the inclusion of unqualified authors^15,21 and the omission of collaborators who effectively contributed to the research.^2,16 On the other hand, three articles mentioned the presence of authors who did not participate significantly in the development of the study or in the writing of the manuscript.^27,33,34 In addition, disputes and disagreements about authorship were reported in different contexts,^2,26,28 pointed out as one of the main sources of unethical pressures in the academic environment.^23,35,36 One study indicated that the order of authorship was defined by the first author based on the level of contribution of each participant.³¹

Honorary authorship — characterized by attribution of authorship to individuals who did not substantially contribute to the study — was described as a recurrent practice in several articles.^{2,12,13,31–34,37}

Only two studies reported that the participants denied having committed or known colleagues who had practiced scientific misconduct.^35,36 However, even in these cases, the interviewees reported having been subjected to unethical pressures related to the order of authorship, to the methodological design or to the presentation of the results, without clarifying what attitudes they had taken in these situations. A third study also addressed similar reports³⁸

Summary of the results

Twenty-three studies were included in the present quantitative synthesis. A total of 17,919 responses were analyzed, corresponding to the estimates extracted from studies evaluating distinct types of scientific misconduct. Among the scientific misconduct behaviors evaluated, honorary authorship presented the highest estimated prevalence, with 32% [95% CI = 21%–44%], followed by plagiarism (28%, 95% CI = 12%– 47%), data falsification (12%, 95% CI = 6%–19%) and data fabrication (11%, 95% CI = 6%–19%). Methodology manipulation was the least frequent behavior, with a prevalence of 4% [95% CI = 2%–7%].

High heterogeneity was observed for most outcomes (I² ≥ 95%), except for methodological manipulation (I² = 62%), which indicates substantial variation among the studies (Table 1).

Table 1.

Characteristics of the included studies (n = 29).

STUDY, author/year (country)	STUDY design	Sample size	Qualification of the interviewees	Results
30. Abdelnaby et al., 2024, Egypt	Cross-sectional	283	Graduate students and faculty members	Between 16% and 18% of the participants in this study reported having committed plagiarism.
14. Adeleye et al. 2012, Nigeria	Cross-sectional	132	Faculty members and consultants	54.6% of the respondents admitted to at least one research misconduct, with inadequate record-keeping being the most frequent, followed by data fabrication, falsification, and plagiarism, respectively.
31. Aldughmi et al. 2021, Jordan	Cross-sectional	148	Full professors and assistant professors	71.7% of the participants were aware of the inclusion of honorary authorship in research.
32. Alshogran et al., 2018 - Jordan	Cross-sectional	272	Faculty members	One third of the respondents reported disputes regarding authorship, with 22.4% believing they were excluded from deserved authorship and 16.5% reporting having granted authorship as a gift.
15. Artino et al., 2019, USA	Cross-sectional	590	Researchers	90.3% of the respondents reported at least one research misconduct, with 60.6% stating that they added unqualified authors to the authorship. 49.5% cited articles in their research that they had not read, and 49.4% cited articles to please editors or reviewers. 5.5% committed plagiarism, 5.3% altered results, and 2.4% fabricated data.
11. Asman et al., 2017 Israel	Cross-sectional	151	Researchers	26.2% of the participants showed a tendency to selectively report or omit data. 15.5% admitted to fabricating data, and 34.2% reported awareness of research misconduct in their workplace.
16. Bebeu et al., 1996, USA	Cross-sectional	76	Professors	Data falsification was the research misconduct reported by 30% of the respondents.
35. Bredhal Jensen et al., 2018, Denmark	Cross-sectional	165	Doctoral students	22% of the respondents stated that they were exposed to unethical pressures related to authorship. However, none admitted to having committed research misconduct.
34. Chen et al., 2024, China	Cross-sectional	6200	Graduate students in Residency programs	50.6% reported multiple submissions of articles, with 46.7% referring to duplicate publication. The buying and selling of articles had a prevalence of 46.7%. 48.7% reported plagiarism, while falsification of research data, materials, literature, or notes, and fabrication of research results totaled 49.0%. Listing false names as authors or failing to list authors who contributed to the research and manuscript accounted for 47.6%.
17. Dhingra et al., 2014, India	Cross-sectional	155	Researchers	20.6% of the respondents reported knowing that a researcher had published duplicate studies with minimal changes. Plagiarism was observed by 53.5% of the respondents, and 32.9% reported knowing that colleagues had split their studies to increase the number of publications.
18. Geggie et al., 2001, United Kingdom	Cross-sectional	194	Consultants	In total, 55.7% of the respondents observed some form of research misconduct, but only 5.7% admitted to having committed it in the past.
19. Haberman et al., 2010, India	Cross-sectional	266	Research coordinators	17% reported awareness of data fabrication in research.
12. Hadji et al., 2016, Iran	Cross-sectional	297	Researchers	The prevalence of data fabrication was 4.15%, plagiarism was 4.90%, and honorary authorship, falsification methods, and duplicate publication were 18.10%, 12.65%, and 5.40%, respectively.
36. Hofmann et al., 2013 Sweden	Cross-sectional	189	Doctoral students	None of the respondents reported having fabricated, falsified, or plagiarized. 11% of the respondents reported experiencing unethical pressure regarding the order of authorship.
20. Hofmann et al., 2020, Sweden	Cross-sectional	285	Doctoral students	10% of the participants reported having altered research data. 10% admitted to plagiarism, and 36% added one or more authors who did not qualify for authorship.
38. Hofmann et al., 2023, Sweden	Cross-sectional	94	Doctoral students	In total, 1.5% of the respondents reported feeling pressured to engage in scientific misconduct.
22. Ismail et al., 2018 - Iraq	Cross-sectional	400	Undergraduate students	81.1% of the participants reported having committed plagiarism.
37. Khot et al., 2024, India	Cross-sectional	146	Graduate students and faculty members	30.3% reported premature publication in predatory journals, and 33.19% cited articles and/or materials they had not read. Plagiarism and duplication of research work accounted for 18.5%. Listing supervisors’ names as authors without their active role in the research was reported by 10.2%, and adding spouses’ and friends’ names as authors without significant contribution was reported by 7.5%.
23. Nilstun et al., 2009, Sweden	Cross-sectional	134	Doctoral students	None of the respondents reported feeling pressured to fabricate data or plagiarize in research. 5% reported experiencing pressure to falsify data, with 3% admitting to having done so.
24. Nortes et al., 2023, Switzerland	Cross-sectional	79	Undergraduate and master’s students	About 13% of the participants admitted to having committed plagiarism.
13. O’Brien et al., 2009, Canadá	Cross-sectional	127	Researchers	51.6% of the respondents reported having been cited as honorary authors in articles.
21. Iloh et al., 2020, Nigeria	Cross-sectional	210	Researchers	All respondents stated they were aware of plagiarism. 97.2% were knowledgeable about data fabrication and falsification, and 86.1% about honorary authorship.
26. Pryor et al., 2006, USA	Cross-sectional	1645	Researchers	The most frequently perceived practice was protocol violations, reported by 43% of the respondents.
2. Rohwer et al., 2017, South Africa	Cross-sectional	199	Researchers	24% stated that they had invited authors who did not effectively participate in the research, and 57% reported that they had not committed this practice but were aware that it occurred. 2% omitted an author who had contributed to the research.
25. Ogundele so et al., 2019, Nigeria	Cross-sectional	94	Professors	The most common research misconduct practice reported by the respondents was plagiarism, mentioned by 25% of them. Data fabrication and falsification were reported by 22% and 21%, respectively. Honorary authorship was reported by 20% of the participants.
27. Tijdink, 2016, Netherlands	Cross-sectional	535	Researchers	Nearly 60% of the participants admitted to adding authors who had not made any significant contribution to the research. Practices such as data fabrication and falsification were less common, but nearly 5% admitted to selectively deleting data to confirm a hypothesis.
33. Wasseem et al., 2022, Pakistan	Cross-sectional	95	Researchers	27.9% added author(s) who did not contribute to the research. The majority of participants (65.26%) reported not submitting a manuscript without the consent of all authors.
28. Were et al., 2020, Kenya	Cross-sectional	577	Researchers	The most frequently reported misconducts were data fabrication 24%, data falsification 22%, and plagiarism 22%. Authorship disputes were reported by 33% of respondents.
29. Yu et al., 2020, China	Cross-sectional	217	Researchers	17.51% admitted to having committed data fabrication, falsification, or plagiarism.

Subgroup analyses based on the country of origin of the studies and the professional area of the evaluated samples did not explain the substantial heterogeneity observed in the pooled estimates (Table 2).

Table 2.

Pooled estimates stratified by study country and sample area.

	Variable		Number of studies included	Sample size	PP [CI95%]	I²
Honorary authorship	Country	Jordan	2	420	26% [9% - 48%]	95%
		Iran	1	297	18% [14% - 23%]	-
		Sweden	1	94	71% [61% - 80%]	-
		India	1	146	10% [6% - 16%]	-
		Canada	1	127	51% [42% - 60%]	-
		South Africa	1	199	24% [18% - 30%]	-
		Nigeria	1	94	19% [12% - 29%]	-
		Netherlands	1	535	60% [56% - 64%]	-
		Pakistan	1	95	27% [19% - 37%]	-
		China	1	217	29% [24% - 36%]	-
	Area	Health sciences in general	6	1260	28% [19% - 39%]	93%
		Medicine	3	775	45% [11% - 83%]	99%
		Medicine and dentistry	1	94	19% [12% - 29%]	-
		Dentistry	1	95	27% [19% - 38%]	-
Data fabrication	Country	Nigeria	2	226	17% [ 11% - 25%]	44%
		United States of America	1	590	2% [1% - 4%]	-
		Israel	1	151	15% [10% - 22%]	-
		India	1	266	17% [13% - 22%]	-
		Iran	1	297	4% [2% - 7%]	-
		Sweden	3	608	8% [0% - 29%]	98%
		Kenya	1	577	24% [20% - 28%]	-
	Area	Health sciences in general	2	1464	8% [0% - 24%]	99%
		Medicine	3	608	8% [0% - 29%]	98%
		Medicine and dentistry	2	226	17% [11% - 25%]	44%
		Nursing	3	417	17% [13% - 20%]	0%
Data falsification	Country	Nigeria	2	226	14% [6% - 26%]	79%
		Israel	1	151	26% [19% - 33%]	-
		United States of America	1	76	30% [21% - 42%]	-
		India	1	266	6% [4% - 10%]	-
		Iran	1	297	12% [9% - 17%]	-
		Sweden	2	323	2% [0% - 9%]	86%
		Netherlands	1	535	5% [3% - 7%]	-
		Kenya	1	577	22% [19% - 26%]	-
	Area	Health sciences in general	2	874	17% [9% - 27%]	92%
		Medicine	3	858	3% [1% - 7%]	83%
		Medicine and dentistry	2	226	14% [6% - 26%]	79%
		Dentistry	1	76	30% [21% - 42%]	-
		Nursing	2	417	15% [1% - 38%]	97%
Methodology manipulation	Country	Nigeria	1	132	1% [0% - 4%]	-
		United States of America	1	590	5% [3% - 7%]	-
		Sweden	2	379	5% [3% - 8%]	0%
	Area	Health sciences in general	590	1	5% [3% - 7%]	-
		Medicine and dentistry	132	1	1% [0% - 4%]	-
		Medicine	379	2	5% [3% - 8%]	0%
Plagiarism	Country	Egypt	1	283	18% [14% - 23%]	-
		Nigeria	3	436	47% [0% - 100%]	100%
		United States of America	2	666	25% [0% - 80%]	99%
		China	1	6200	49% [47% - 50%]	-
		India	2	301	35% [7% - 71%]	98%
		Iran	1	297	5% [2% - 8%]	-
		Sweden	1	189	1% [0% - 3%]	-
		Iraq	1	400	54% [49% - 59%]	-
		Switzerland	1	79	13% [6% - 22%]	-
		Kenya	1	577	22% [19% - 26%]	-
	Area	Health sciences in general	4	1747	11% [4% - 22%]	97%
		Medicine and dentistry	2	226	13% [0% - 38%]	95%
		Medicine	5	6900	45% [7% - 87%]	100%
		Dentistry	1	76	54% [42% - 65%]	-
		Nursing	1	79	13% [7% - 22%]	-
		Medicine and nursing	1	400	54% [49% - 59%]	-

Meta-regression analyses were feasible only for study sample size across all domains assessed in the meta-analysis, and for mean age exclusively in the plagiarism domain, due to the insufficient number of studies reporting this variable for the remaining domains, which predominantly presented age data in intervals rather than as continuous measures. None of the examined covariates showed a statistically significant association with the effect estimates (p > 0.05) (Table 3).

Table 3.

Meta-regression results based on study sample size and the mean age of the evaluated population.

Domain	Predictor variable	Number of studies included	p-value*
Honorary authorship	Sample size	11	0.639
Data fabrication	Sample size	10	0.583
Data falsification	Sample size	10	0.718
Plagiarism	Sample size	14	0.598
Plagiarism	Mean age of the sample	8	0.609

*p-value of the moderator test, indicating whether the variable included in the meta-regression is associated with the differences in effect observed across studies.

Reporting bias

Asymmetry was not observed in the funnel graph (p > 0.05), and there was no evidence of publication bias among the studies included in the meta-analysis. This absence of asymmetry suggests that the results were not influenced by the non-disclosure of studies with smaller or non-significant estimates (Appendix 3).

Discussion

Scientific misconduct refers to unethical behaviors or practices in conducting research, which compromise the integrity and credibility of the results obtained. Such practices undermine confidence in science by disseminating conclusions, often mistaken, that negatively impact the advancement of knowledge and its practical application in health care.^39,40 In addition, they result in the waste of financial resources for scientific production.⁴¹ The present study aimed to evaluate the prevalence of different forms of scientific misconduct in the health area through a systematic review with meta-analysis. The results indicated that practices related to improper authorship and plagiarism are more frequent among the studies analyzed, while behaviors such as falsification, data fabrication and methodological manipulation were less frequently reported. These findings show the persistence of unethical conduct in the scientific environment and the need for institutional actions aimed at promoting integrity in research.

It is the responsibility of researchers and institutions to ensure reliability and ethics in conducting research.⁴² However, the pressure for academic productivity and the appreciation of the number of publications have led to the prioritization of institutional partnerships and strategic co-authorships, even when lacking effective contribution.⁴³ This scenario favors the dissemination of inappropriate practices, such as honorary authorship — the most prevalent misconduct identified in this review. The absence of objective criteria for attribution of authorship represents a concrete threat to scientific integrity,⁴⁴ being reinforced by institutional policies that use publication rankings as the main criterion for hiring and promoting teachers, without effectively considering their individual contributions in works with multiple authors.⁴⁵

Institutional pressure to publish is one of the most cited factors as a cause of questionable conduct in research^45–47 affecting mainly those researchers who feel the highest demand, who tend to admit greater involvement in these practices.^27,48 On the other hand, studies show that, even under the guarantee of anonymity, many researchers omit the practice of scientific infractions,¹⁵ which suggests a significant underreporting of the actual prevalence of these behaviors.⁴⁹

With the increase in scientific production in the health area, there was also a significant growth in the number of retractions of articles, most of them resulting from misconduct.⁵⁰ A study conducted in China — a country with a high volume of scientific publications and an estimated prevalence of misconduct at 40%⁵¹ — revealed, through semi-structured interviews, that participants had limited understanding of the subject, recognizing mainly the fabrication, falsification of data and plagiarism as synonyms of scientific misconduct.⁵² The findings of this review coincide with those of this study, indicating that such practices, although frequent, are often not recognized as ethically unacceptable by the authors themselves.

The increasingly frequent use of technology as a tool in the academic environment has contributed to the emergence of new forms of ethical infractions, called by some authors “variants of cheating” in research.^39,53 Tools based on artificial intelligence, such as chatbots — among them ChatGPT —, although useful to assist in academic tasks, do not replace the guidance of specialized professionals, whose performance confers credibility to research through judgments based on technical knowledge and consolidated experience. The ability of artificial intelligence to rewrite excerpts from other authors with similar words should not be considered an acceptable practice in scientific production, because, regardless of the form employed, it is a form disguised as plagiarism.^54,55

Among the misconducts analyzed in this review, plagiarism is the only practice subject to legal sanctions. However, there is consensus in the literature that other offenses should also be subject to equivalent punitive criteria.⁴⁵ Faced with the increasing spread of unethical behaviors, scientific journals began to demand, at the time of submission of articles, a detailed statement of the contribution of each author involved.¹⁵

Limitations and future research

Some limitations should be considered when interpreting the results of this review. First, the pooled estimates were based on self-reported data, which introduces a risk of response bias. Scientific misconduct is a sensitive topic and, even under conditions of anonymity, participants may omit personal or colleagues’ behaviors, leading to underreporting and consequently to underestimation of the true prevalence. Future studies could combine self-reports with objective indicators, such as plagiarism detection tools, retraction databases or analysis of publication metadata.

Second, the included studies differed substantially in terms of sample profiles (students, researchers and health professionals), institutional contexts and methods used for data collection and reporting. This variability likely contributed to the heterogeneity observed in the pooled estimates. Although we applied a random-effects model, which is recommended when true effects are expected to vary across studies, subgroup analyses and meta-regression were conducted using the available study-level variables. However, only study area, country, sample size, and mean age of the sample could be analyzed, and these variables were insufficient to explain the observed heterogeneity. Future studies using standardized instruments and uniform reporting would allow for more refined comparative analyses.

Third, although no asymmetry was detected in the funnel plot, suggesting absence of publication bias among the included studies, this result should be interpreted cautiously. Funnel plot–based tests only assess bias among studies with available quantitative data and do not fully capture other forms of reporting bias, such as selective non-publication of surveys, theses, or studies that did not report prevalence estimates in sufficient detail. This is a limitation inherent to the available evidence rather than to the review process itself. Future research would benefit from standardized reporting, protocol registration, and greater accessibility of primary data.

Another limitation is the substantial heterogeneity observed across the pooled estimates. This variability in professional field, institutional setting, and survey focus may contribute to the observed between-study heterogeneity. High I² values are common in meta-analyses of prevalence due to the sensitivity of proportional data and the intrinsic variability between study contexts.⁵⁶ Although subgroup analyses and meta-regression were conducted using the available study-level variables, including professional field, country, sample size, and mean age, these analyses were insufficient to explain the observed heterogeneity. This limitation reflects the restricted availability, inconsistency, and limited variability of reported covariates across studies. Therefore, although a random-effects model was applied to account for between-study variability, the influence of potential effect modifiers could not be further examined.

Despite these limitations, the findings provide an important overview of the problem and can support the development of educational, institutional and regulatory strategies to strengthen research integrity.

Conclusion

This systematic review and meta-analysis confirmed that scientific misconduct remains a persistent issue in health research, particularly in practices related to authorship and intellectual originality. Honorary authorship was reported by about one in three participants, followed by plagiarism, while data falsification and fabrication, though less frequent, still affected roughly one in eight to nine researchers. Methodological manipulation was less common but not absent. Despite heterogeneity among studies and potential underreporting, the findings consistently revealed structural ethical weaknesses within the scientific environment. The evidence highlights the need for continuous institutional efforts to strengthen research integrity through transparent authorship criteria, mandatory ethics education, and effective mechanisms for accountability and transparency. By including diverse professional profiles and a broader range of inappropriate practices, this study provides updated estimates that can inform policies to promote integrity, prevent misconduct, and foster a culture of responsibility in health research.

Supplemental material

Supplemental Material - Scientific misconduct in health research: A systematic review and meta-analysis

Supplemental Material for Scientific misconduct in health research: A systematic review and meta-analysis by Paloma Alves Miquilussi, Rayane Délcia da Silva, Glória Maria Nogueira Cortz Ravazzi, Karinna Verissimo Meira Taveira, Samara Fernandes da Silva Souza, Bianca Simone Zeigelboim, Diulia Pereira Bubna, Bianca Marques de Mattos de Araujo, Rosane Sampaio Santos, Cristiano Miranda de Araújo in Research Methods in Medicine & Health Sciences

Supplemental material

Supplemental Material - Scientific misconduct in health research: A systematic review and meta-analysis

Supplemental material

Supplemental Material - Scientific misconduct in health research: A systematic review and meta-analysis

Footnotes

ORCID iD

Paloma Alves Miquilussi

Consent to participate

This article does not involve human participants, and therefore informed consent was not required.

Author’s contribuitions

Study design: Samara Fernandes da Silva Souza, Bianca Marques de Mattos de Araujo.

Data collection: Paloma Alves Miquilussi, Rayane Délcia da Silva.

Data analysis: Paloma Alves Miquilussi, Rayane Délcia da Silva, Glória Maria Nogueira Cortz Ravazzi.

Manuscript writing: Paloma Alves Miquilussi, Rayane Délcia da Silva, Glória Maria Nogueira Cortz Ravazzi.

Manuscript revision: Rosane Sampaio Santos, Bianca Simone Zeigelboim, Cristiano Miranda de Araújo, Karinna Verissimo Meira Taveira.

Study supervision: Rosane Sampaio Santos, Cristiano Miranda de Araújo, Bianca Simone Zeigelboim

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental material

Supplemental material for this article is available online.

References

Tomasevicius

Morais

AMF

Barraviera

, et al. Diretrizes do CSE para promover integridade em publicações de periódicos científicos: atualização de 2012. São Paulo: Associação Brasileira de Editores Científicos (ABEC Brasil); 2017.p. 58-93.

Rohwer

Young

Wager

, et al. Authorship, plagiarism and conflict of interest: views and practices from low/middle-income country health researchers. BMJ Open 2017; 7: e018467. https://doi.org/10.1136/bmjopen-2017-018467

Coury HJJBJoPT . Research and scientific publication integrity. SciELO Brasil, 2012. v-vi.

Radiology CoPEJD . Committee on Publication Ethics (COPE). Guidelines on good publication practice. Dentomaxillofac Radiol. 2000; 29: 195–200. https://doi.org/10.1038/sj/dmfr/4600534

Phogat

Manjunath

Sabbarwal

, et al. Misconduct in biomedical research: a meta-analysis and systematic review. J Int Soc Prev Community Dent 2023; 13: 185–193. https://doi.org/10.4103/jispcd.JISPCD_220_22

Mousavi

Abdollahi

. A review of the current concerns about misconduct in medical sciences publications and the consequences. Daru: Journal of Faculty of Pharmacy, Tehran University of Medical Sciences 2020; 28: 359–369. https://doi.org/10.1007/s40199-020-00332-1

Stern

Casadevall

Steen

, et al. Financial costs and personal consequences of research misconduct resulting in retracted publications. eLife 2014; 3: e02956. https://doi.org/10.7554/eLife.02956

Page

McKenzie

Bossuyt

, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. 2021; 372: n71. https://doi.org/10.1136/bmj.n71

Klugar

. A protocol is essential for a systematic review as randomization is for randomized controlled trials. JBI Database System Rev Implement Rep 2016; 14: 1–2. https://doi.org/10.11124/JBISRIR-2016-003008

10.

Aromataris

Lockwood

Porritt

, et al. JBI Manual for Evidence Synthesis. JBI; 2024. Available from: https://synthesismanual.jbi.global.

11.

Asman

Melnikov

Barnoy

, et al. Experiences, behaviors, and perceptions of registered nurses regarding research ethics and misconduct. Nurs Ethics 2019; 26: 859–869. https://doi.org/10.1177/0969733017727152

12.

Hadji

Asghari

Yunesian

, et al. Assessing the prevalence of publication misconduct among Iranian authors using a double list experiment. Iran J Public Health 2016; 45: 897–904.

13.

O'Brien

Baerlocher

Newton

, et al.

Honorary coauthorship: does it matter?

2009; 60: 231–236.

14.

Adeleye

Adebamowo

. Factors associated with research wrongdoing in Nigeria. J Empir Res Hum Res Ethics 2012; 7: 15–24.

15.

Artino

Jr Driessen

Maggio

. Ethical shades of gray: international frequency of scientific misconduct and questionable research practices in health professions education. Acad Med 2019; 94: 76–84. https://doi.org/10.1097/ACM.0000000000002412

16.

Bebeau

Davis

. Survey of ethical issues in dental research. J Dent Res 1996; 75: 845–855. https://doi.org/10.1177/00220345960750021901

17.

Dhingra

Mishra

. Publication misconduct among medical professionals in India. Indian J Med Ethics 2014; 11: 104–107. https://doi.org/10.20529/IJME.2014.026

18.

Geggie

. A survey of newly appointed consultants' attitudes towards research fraud. J Med Ethics 2001; 27: 344–346.

19.

Habermann

Broome

Pryor

, et al. Research coordinators' experiences with scientific misconduct and research integrity. Nurs Res 2010; 59: 51–57. https://doi.org/10.1097/NNR.0b013e3181c3b9f2

20.

Hofmann

Bredahl Jensen

Eriksen

, et al. Research integrity among PhD students at the faculty of medicine: a comparison of three Scandinavian universities. J Empir Res Hum Res Ethics 2020; 15: 320–329. https://doi.org/10.1177/1556264620929230

21.

Iloh

Amadi

Iro

, et al. Attitude, practice orientation, benefits and barriers towards health research and publications among medical practitioners in Abia State. Nigeria: A cross-sectional study 2020; 23: 129–137.

22.

Ismail

. Perceptions of plagiarism among medical and nursing students in Erbil, Iraq. Sultan Qaboos Univ Med J 2018; 18: e196.

23.

Nilstun

Löfmark

Lundqvist

. Scientific dishonesty—questionnaire to doctoral students in Sweden. J Med Ethics 2010; 36: 315–318.

24.

Nortes

Fierz

Goddiksen

, et al. Academic integrity among nursing students: a survey of knowledge and behavior. Nurs Ethics. 2024 Jun;31(4):553–571. https://doi.org/10.1177/09697330231200568

25.

Ogundele

Kuyinu

Salisu

, et al. Knowledge of research and publication ethics in a Nigerian medical college. Ann Health Res 2019; 5: 165–173. https://doi.org/10.30442/ahr.0502-18-48

26.

Pryor

Habermann

Broome

. Scientific misconduct from the perspective of research coordinators: a national survey. 2007; 33: 365–369.

27.

Tijdink

Bouter

Veldkamp

, et al. Personality traits are associated with research misbehavior in Dutch scientists: a cross-sectional study. PLoS One 2016; 11: e0163251. https://doi.org/10.1371/journal.pone.0163251

28.

Were

Kiplagat

Kaguiri

, et al. Institutional capacity to prevent and manage research misconduct: perspectives from Kenyan research regulators. Res Integr Peer Rev 2023; 8: 8. https://doi.org/10.1186/s41073-023-00132-6

29.

Miao

Liu

, et al. Scientific misconduct and associated factors: a survey of researchers in three Chinese tertiary hospitals. Account Res 2021; 28: 95–114. https://doi.org/10.1080/08989621.2020.1809386

30.

Abdelnaby

Abbas

Abdelhafiz

, et al. Knowledge, attitudes, and self-reported practices regarding plagiarism among health researchers in Egypt and Lebanon: a multicenter questionnaire study. Research Ethics. 2024;21(1):17470161241291937. https://doi.org/10.1177/17470161241291937

31.

Aldughmi

Qutaishat

Karasneh

RJS

. Knowledge and perceptions of honorary authorship among health care researchers: online cross-sectional survey data from the middle east. Sci Eng Ethics 2021; 27: 39. https://doi.org/10.1007/s11948-021-00317-6

32.

Alshogran

Al-Delaimy

. Understanding of international committee of medical journal editors authorship criteria among faculty members of pharmacy and other health sciences in Jordan. J Empir Res Hum Res Ethics 2018; 13: 276–284.

33.

Waseem

Awan

Malik

, et al. Questionable research practices among dentists. 2022; 16: 415. https://doi.org/10.53350/pjmhs22169415

34.

Chen

Wang

, et al. Knowledge, attitudes and practices about research misconduct among medical residents in southwest China: a cross-sectional study. BMC Med Educ 2024; 24: 284. https://doi.org/10.1186/s12909-024-05277-6

35.

Bredahl

Kyvik

Leth-Larsen

, et al. Research integrity among PhD students within clinical research at the University of Southern Denmark. 2018; 65: A5469.

36.

Hofmann

Myhr

Holm

SJB

. Scientific dishonesty—a nationwide survey of doctoral students in Norway. BMC Med Ethics 2013; 14: 1–9. https://doi.org/10.1186/1472-6939-14-3

37.

Khot

Chindhalore

Naikwadi

AJC

. Knowledge, attitude, and practices about research integrity and scientific misconduct among the faculty and medical postgraduates working in medical colleges in north Karnataka and central India: a cross-sectional online survey. Cureus 2024; 16: e59200. https://doi.org/10.7759/cureus.59200

38.

Hofmann

Thoresen

Holm

. Research Integrity Attitudes and Behaviors are difficult to alter: results from a ten year Follow-up study in Norway. J Empir Res Hum Res Ethics. 2023; 18: 50–57.

39.

Birks

Smithson

Antney

, et al. Exploring the paradox: a cross-sectional study of academic dishonesty among Australian nursing students. Nurse Educ Today 2018; 65: 96–101. https://doi.org/10.1016/j.nedt.2018.02.040

40.

Bülow

Helgesson

. Health care and philosophy. Criminalization of scientific misconduct. Med Health Care Philos 2019 Jun;22: 245–252. https://doi.org/10.1007/s11019-018-9865-7

41.

Bordewijk

van Eekelen

, et al. Methods to assess research misconduct in health-related research: a scoping review. J Clin Epidemiol 2021; 136: 189–202. https://doi.org/10.1016/j.jclinepi.2021.05.012

42.

D’Souza

Sade

Moffatt-Bruce

, et al. The many facets of research integrity: what can we do to ensure it? J Thorac Cardiovasc Surg. 2020; 160: 730–733.

43.

Breet

Botha

Horn

, et al. Academic and scientific authorship practices: a survey among South African researchers. J Empir Res Hum Res Ethics 2018; 13: 412–420. https://doi.org/10.1177/1556264618789253

44.

Pruschak

Hopp

. And the credit goes to…-ghost and honorary authorship among social scientists. PLOS ONE. 2022; 17: e0267312.

45.

Arjuman

Chakraborty

. Plagiarism and authorship credit. Indian J Public Health. 2024; 68: 318–323.

46.

Dinis-Oliveira

Magalhães

TJF

. The inherent drawbacks of the pressure to publish in health sciences: good or bad science. F1000Res 2015; 4: 419. https://doi.org/10.12688/f1000research.6809.2

47.

Maggio

Dong

Driessen

, et al. Factors associated with scientific misconduct and questionable research practices in health professions education. Perspect Med Educ 2019; 8: 74–82. https://doi.org/10.1007/s40037-019-0501-x

48.

Fanelli

Costas

Larivière

. Misconduct policies, academic culture and career stage, not gender or pressures to publish, affect scientific integrity. PLOS ONE. 2015; 10: e0127556. https://doi.org/10.1371/journal.pone.0127556

49.

Laar

Redman

Ferguson

, et al. Institutional approaches to research integrity in Ghana. Sci Eng Ethics 2020; 26: 3037–3052. https://doi.org/10.1007/s11948-020-00257-7

50.

Stavale

Pupovac

Ferreira

, et al. Research integrity guidelines in the academic environment: the context of Brazilian institutions with retracted publications in health and life sciences. Front Res Metr Anal 2022; 7: 991836. https://doi.org/10.3389/frma.2022.991836

51.

Liao

Q-J

Zhang

Y-Y

Fan

Y-C

, et al. Perceptions of Chinese biomedical researchers towards academic misconduct: a comparison between 2015 and 2010. Sci Eng Ethics 2018; 24: 629–645. https://doi.org/10.1007/s11948-017-9913-3

52.

Nemery

Dierickx

. Perceptions of research integrity and the Chinese situation: In-depth interviews with Chinese biomedical researchers in Europe. Account Res. 2019; 26: 405–426.

53.

Elali

Rachid

. AI-generated research paper fabrication and plagiarism in the scientific community. Patterns (N Y) 2023; 4: 100706. https://doi.org/10.1016/j.patter.2023.100706

54.

Salvagno

Taccone

Gerli

. Can artificial intelligence help for scientific writing? Crit Care. 2023; 27: 75.

55.

Boscardin

Gin

Golde

, et al. ChatGPT and generative artificial intelligence for medical education: potential impact and opportunity. Acad Med 2024; 99: 22–27. https://doi.org/10.1097/ACM.0000000000005439

56.

Migliavaca

Stein

Colpani

, et al. Meta-analysis of prevalence: I(2) statistic and how to deal with heterogeneity. Research synthesis methods 2022; 13: 363–367. https://doi.org/10.1002/jrsm.1547

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.20 MB

0.22 MB

0.24 MB