# BILSER: Research-integrity policy Standards for trustworthy research, participant protection, responsible publication and impartial handling of concerns. Status: Prepared for adoption Document: GOV-04; version 1.0 Prepared / sources checked: 18 October 2026 Responsible office: Scientific Director and an independent integrity lead, upon appointment Adoption authority: Competent governing body under the valid constitution Effective date: Not yet recorded Review: Annually after adoption and when requirements change This version is prepared for institutional review and adoption. An approving resolution, effective date, and signed adoption record have not been recorded. Publication does not amend a registered constitution or establish a committee appointment. Bangladesh legal review and any required regulatory filings must precede implementation. ## 1. Scope and personal responsibility This proposed policy covers research designed, funded, managed, hosted or represented by BILSER and applies to every participating researcher, supervisor, trainee, volunteer and contractor. Collaborations must identify the responsible institutions and agree oversight in writing. Researchers remain accountable for their own contributions. BILSER's proposed standards require truthful records, rigorous methods, fair treatment, accountable stewardship and a research culture in which concerns can be raised. Scientific disagreement and honest error require correction or discussion; they are not automatically misconduct. These standards take effect through formal adoption and appropriate agreements. ## 2. Before research begins - Approve a scientifically justified protocol, responsible investigator, budget, supervision plan, data-management plan and documented risk assessment before activity starts. Obtain all activity-specific permissions and partner agreements. - Research involving people or identifiable human material or data requires a prospective decision from the competent independent ethics committee through the applicable Bangladesh pathway. An exemption or waiver must be documented by the competent review body, not assumed by the investigator. Confirm BMRC requirements for the study and collaboration. - Use meaningful informed consent in a language participants understand, protection for vulnerable participants, fair recruitment, privacy safeguards, appropriate care and arrangements for research-related harm. Any consent waiver, change or secondary use needs the applicable ethics determination before implementation. - Register clinical trials prospectively in an appropriate publicly accessible registry before the first participant is recruited and meet the applicable regulator, ethics-committee and journal requirements. Do not describe registration as regulatory approval. - Secure applicable animal-welfare, environmental, collection, laboratory, biosafety, biosecurity and material-transfer approvals before the corresponding work. Assess misuse risks proportionately and escalate unresolved high-risk work to competent independent review. ## 3. Design, conduct and reproducibility - Choose methods and sample sizes justified by the research question, use appropriate controls and analysis, and record protocol versions and deviations. Pre-register hypotheses and analysis plans where appropriate to the design. - Preserve original observations, laboratory records, instrument outputs, code and analytic decisions with dated provenance and access controls. Do not overwrite raw data, conceal inconvenient results, manipulate images misleadingly or invent observations. - Describe exclusions, missing data, uncertainty, limitations and exploratory analyses transparently. Arrange proportionate independent checking of important analyses, figures and conclusions before submission. - Stop or modify activity where an immediate safety concern arises, protect participants and notify the responsible ethics body, sponsor or authority through the required process. Obtain approval for material amendments before implementing them except necessary immediate protective action allowed by the applicable framework. ## 4. Data, materials and confidentiality Assign a data custodian and document collection, lawful use, access, security, backup, retention, sharing and disposal before collection begins. Separate direct identifiers where practicable, use controlled access and protected backups, and assess re-identification risks before release. Use data and material agreements for external access and transfers. The proposed institutional default is to retain the research record for at least ten years after final publication or study closure, whichever is later, subject to the approved protocol and applicable law. Retain identifiable information only for the lawful, justified period established in the approved schedule. Apply longer lawful funder or regulatory requirements and legal holds; resolve incompatible privacy or consent limits before collection rather than retaining data indefinitely. Share research outputs as openly as consent, law, contracts, participant protection, legitimate intellectual-property arrangements and security permit. If unrestricted sharing is inappropriate, provide metadata and a controlled-access explanation where feasible. Record and contain suspected breaches promptly and obtain advice on required notifications; this policy does not invent a universal statutory reporting deadline. ## 5. Authorship, attribution and the public record Agree contributor roles early and revisit them as work develops. For medical publications, an author must contribute substantively to the work, take part in drafting or substantive intellectual review, approve the final manuscript and accept accountability. Give eligible contributors a fair opportunity to meet those responsibilities. Apply the destination journal's requirements and record contributions transparently. Do not grant honorary, purchased or ghost authorship. Acknowledge other contributions with permission. Declare funding and relevant interests. Preserve original affiliations on earlier publications; a later appointment does not turn previous work into a BILSER institutional output. Publish only consented, accurate researcher profiles and appointments. Report findings accurately, including negative or inconclusive results where appropriate. Avoid duplicate submission, plagiarism, misleading selective reporting and unsupported claims. Correct material errors promptly with the journal or repository, and cooperate with justified correction, expression-of-concern and retraction processes. ## 6. AI, peer review and collaboration - Human researchers remain responsible for AI-assisted output. Validate generated text, code, analysis, citations and images against evidence; document tools and material use in the research record and disclose them as the journal, funder or institution requires. AI tools cannot be named as authors. - Do not upload identifiable participant data, confidential proposals, unpublished manuscripts or restricted materials into an external AI service without an approved legal, ethical, contractual and security assessment and appropriate authorization. Check retention, training use, access and cross-border processing before use. - Peer reviewers must declare relevant interests, maintain confidentiality and give fair, constructive assessments. Do not use privileged information for personal advantage or delegate confidential review to an AI service without express authorization. - Collaboration agreements must allocate responsibilities for approvals, data stewardship, intellectual property, contributions, dissemination and complaints. Include local partners fairly and do not claim facilities or institutional endorsement without a valid agreement. ## 7. Raising a concern safely Report suspected fabrication, falsification, plagiarism, serious ethics breaches, deliberate non-disclosure, misuse of funds or data, unsafe research or retaliation. Provide a factual description, dates and the location of relevant evidence where possible. Anonymous concerns may be assessed when sufficiently specific; an unsubstantiated good-faith report is not misconduct. An independent integrity contact must be appointed before this framework becomes operational. Until then, use an unconflicted published leader to request a safe reporting channel, without initially sending sensitive evidence. A concern about the Chairperson should go to an unconflicted Scientific Director. If leadership is conflicted, unavailable or unsafe, approach the relevant host institution, ethics committee, funder or competent authority directly. The general mailbox is not a confidential whistleblowing service. Limit disclosure to those who need the information and use secure case records. Explain the limits of confidentiality, including required legal or regulatory disclosures. No promise of absolute anonymity is made. Protect reporters, witnesses and respondents against retaliation; protective measures must not presume a finding of misconduct. ## 8. Intake, protection and preliminary assessment - The appointed unconflicted integrity lead acknowledges an identifiable report within five working days where safe, assesses immediate welfare and evidence risks promptly, and records jurisdiction, conflicts and required external notifications. These are proposed service targets, not legal deadlines. - Preserve relevant records lawfully with a documented chain of custody. Use proportionate temporary safeguards for participants, funds and data. Do not destroy records, obstruct lawful reporting or impose a punitive suspension before a fair decision. - Aim to complete an initial assessment within 30 working days: close with reasons, refer a routine dispute or honest error for correction, refer to the proper institution, or commission a formal investigation. Record extensions and inform affected parties where lawful and safe. - Coordinate with host institutions, ethics committees, funders and authorities where their jurisdiction applies. Make urgent or mandatory notifications on the applicable timetable without waiting for the internal process. Do not claim criminal-investigation powers. ## 9. Independent investigation and fair process Unconflicted authorized decision-makers appoint a competent panel with a majority independent of the work and an external chair where internal independence is inadequate. Screen relationships and document terms of reference. Panel members must not be parties, complainants or people who have prejudged the case. Provide a route to raise a reasoned panel-conflict objection. Give the respondent written allegations, the evidence needed for a meaningful response, reasonable time to respond and an opportunity to be heard and assisted by a support person, subject to lawful confidentiality measures. Allow correction of factual inaccuracies in the draft findings. Assess all relevant evidence fairly, including exculpatory material. For institutional findings, use the balance of probabilities unless a binding legal or contractual process requires another standard. For a fabrication, falsification or plagiarism misconduct finding, establish a significant departure from accepted practice committed intentionally, knowingly or recklessly; do not treat honest error or scientific disagreement as such a finding. Other policy breaches require their own stated basis. This internal standard determines no criminal liability. Aim to complete the investigation within 90 working days after panel appointment. Record justified extensions, evidence, findings on each allegation, reasons and recommended action. An unconflicted authorized body makes the outcome decision; anyone materially involved in the allegations or original assessment must not control it. ## 10. Outcomes, appeal and restoration Action must be proportionate, lawful and consistent with contracts: correction, training, supervision, reassignment, restrictions or disciplinary referral as justified. Notify journals, funders, partner institutions, ethics bodies or regulators where required and provide an accurate account without unnecessary personal disclosure. Preserve the decision record and protect research participants. The respondent may appeal within 30 calendar days on material new evidence, procedural unfairness, an unreasonable evidential conclusion or a decision-maker conflict. A reporter may request review of an alleged procedural failure. An independent reviewer not previously involved records a reasoned outcome, with a target of 30 working days or an explained extension. No internal deadline limits statutory rights or lawful external reporting. Where allegations are not substantiated, communicate the outcome to the relevant recipients as appropriate and take reasonable steps to restore reputation and working relationships. Treat deliberately false allegations as a separate matter requiring evidence and fair process; lack of proof alone is not bad faith. ## 11. Training, retention and review - Provide induction before research participation and annual refresher training on ethics, records, authorship, conflicts, AI use, safety and raising concerns. Supervisors record training and provide meaningful oversight. - Assign case-record custody to an unconflicted authorized officer. The proposed case-record retention period is seven years after final closure, subject to lawful privacy limits, longer binding requirements and legal holds. Record access and dispose of records securely when authorized. - Review implementation annually using anonymized findings, response times, training coverage and corrective actions. Reassess after significant incidents, changes in law or new research risks. Publish version changes without exposing confidential cases. ## Sources and applicability Bangladesh law applies where relevant. International benchmarks are not legislation or institutional certification. Institutional procedures and time targets are proposed rules, not statutory deadlines. Confirm the current legal position and required approvals before adoption. - [Bangladesh Medical Research Council](https://bmrcbd.org/) (National research-ethics authority and guidance). Confirm the applicable national ethical-clearance pathway and latest guidance directly with BMRC and the responsible ethics committee. The legacy guideline download was unavailable during this review. - [WMA Declaration of Helsinki, 2024 revision](https://www.wma.net/policies-post/wma-declaration-of-helsinki/) (International ethical benchmark). A benchmark for participant protection in medical research. It does not replace Bangladesh permissions or an independent ethics decision. - [Singapore Statement on Research Integrity](https://www.wcrif.org/guidance/singapore-statement) (International voluntary guidance). A principles-based reference, not legislation or certification of this institution. - [ALLEA European Code of Conduct for Research Integrity, 2023](https://allea.org/code-of-conduct/) (International professional benchmark). A research-integrity framework used here as guidance. European law is not made applicable merely by referencing this code. - [ICMJE recommendations on authors and contributors](https://www.icmje.org/recommendations/browse/roles-and-responsibilities/defining-the-role-of-authors-and-contributors.html) (Publication guidance). Informs authorship, contributor recognition and human responsibility for AI-assisted work. The destination journal's current requirements also apply. - [ICMJE recommendations on disclosure of relationships and activities](https://www.icmje.org/recommendations/browse/roles-and-responsibilities/author-responsibilities--conflicts-of-interest.html) (Publication guidance). Informs transparent publication disclosures and sponsor independence. Institutional disclosure rules below are BILSER proposals.