Data Safety Monitoring Plan (DSMP)

A Data Safety Monitoring Plan (DSMP) is the written document a research team prepares — typically as part of a grant application, protocol, or IRB submission, before a trial opens — that spells out exactly how participant safety data will be monitored for the life of the study. It answers four questions in writing: who reviews safety data, what gets reviewed and how often, how adverse events get reported to the IRB and funding/regulatory agencies, and what would trigger pausing, modifying, or stopping the trial. Every NIH-funded clinical trial is required to have a DSMP under the National Institutes of Health’s 1998 Policy for Data and Safety Monitoring (NOT-98-084), which requires that monitoring be commensurate with the trial’s size, complexity, and degree of risk to participants — but the policy does not require every trial to establish a full independent Data Safety Monitoring Board (DSMB) to satisfy that requirement.

DSMP vs. DSMB: the document vs. one possible mechanism

These two terms are frequently used interchangeably, but they describe different things. A DSMB is a body — an independent group of people who review accumulating trial data at intervals and recommend whether the trial should continue, change, or stop. A DSMP is a document — the written plan that describes the study’s overall safety-monitoring approach, whatever form that monitoring takes. A DSMB is one mechanism a DSMP can specify; it is not the only one, and most DSMPs across the full range of NIH- and industry-sponsored trials do not specify one, because most trials are not large multi-site Phase III studies.

Put differently: every trial needs a DSMP. Not every DSMP needs a DSMB. A trial can have a complete, adequate, funder-approved DSMP that names the principal investigator or an independent safety/medical monitor as the reviewer, with no board convened at all.

What a DSMP must cover

NIH and its institutes and centers (NIAID, NIMH, NIAAA, NIAMS, and others each publish their own DSMP-writing guidance, applying the same underlying policy) converge on the same core elements a DSMP should specify in writing:

• Who is responsible for monitoring. The specific individual or body designated — the PI, an independent safety monitor or designated medical monitor, a safety monitoring committee, or a full DSMB — named by role, not left implicit.
• What will be monitored, and how often. The categories of data under review (adverse events, protocol deviations, enrollment and retention, data quality/completeness) and the review frequency — continuous, monthly, quarterly, or tied to enrollment/event milestones.
• The adverse-event reporting mechanism. How adverse events, especially serious or unexpected ones, get reported up the chain — to the IRB, to the funding agency (NIH), and, for FDA-regulated studies, to the FDA — and on what timeline.
• Stopping rules and thresholds. The pre-specified conditions — a safety signal, a dose-limiting toxicity rate, an interim efficacy or futility boundary — that would trigger pausing enrollment, amending the protocol, or halting the trial.
• Data accuracy and protocol-compliance procedures. How the plan assures the underlying data being monitored is itself accurate and that the protocol is being followed as written.

A protocol section that says only “safety will be monitored throughout the study,” with no reviewer named, no schedule, no reporting mechanism, and no stopping criteria, does not meet this bar — reviewers and IRBs expect each element addressed explicitly, scaled to the study’s actual risk profile.

When a DSMP alone is sufficient vs. when a full DSMB is required

The controlling principle, per NIH’s 1998 policy, is that the level of monitoring described in the DSMP should scale to the trial’s risk, size, and complexity — not that every DSMP must escalate to a board. In practice:

• Small, single-site, lower-risk studies — commonly early-phase, small-sample, or minimal-risk designs — can satisfy the DSMP requirement with monitoring by the principal investigator or by an independent safety monitor/designated medical monitor who is not otherwise part of the study team, reviewing data on a defined schedule with defined stopping criteria. No board is convened; the PI’s or monitor’s review, as described in the DSMP, is the entire monitoring mechanism.
• Multi-site trials involving greater-than-minimal risk to participants, and most notably Phase III trials, generally require a full independent DSMB as the mechanism the DSMP specifies — the added complexity, participant volume, and risk profile of these trials are exactly what an independent board with access to unblinded, by-arm data is designed to catch that a single PI reviewing blinded or site-level data typically cannot.
• Between those two ends, funders and IRBs weigh factors including the population’s vulnerability, the intervention’s known or plausible risk, whether interim efficacy analyses are planned, and prior safety experience with the same or a related intervention — the same factors FDA’s guidance on data monitoring committees uses when recommending (not mandating) a committee for industry-sponsored trials.

Because this is a risk-scaled judgment rather than a fixed rule, some NIH institutes publish a decision aid (for example, NIAID’s DSMP decision tree) to help investigators determine, for their specific study design and risk level, whether a DSMB is expected or whether PI/medical-monitor-level review is adequate. Always check the specific funding institute’s current guidance rather than assuming one threshold applies across all of NIH.

Who reviews safety data when there is no DSMB

When a DSMP does not establish an independent board, the plan must still name a specific reviewer and describe their independence from the study’s day-to-day conduct where feasible. Common arrangements include:

• The principal investigator personally reviews accumulating safety data against the plan’s schedule and stopping rules — appropriate for small, low-risk, single-site studies where the added independence a board provides is not proportionate to the trial’s risk.
• An independent safety monitor or designated medical monitor — a qualified individual, external to the study team, appointed specifically to review safety data on the schedule the DSMP defines — gives the plan a degree of independence between a PI-only review and a full board, without the administrative overhead of convening one.
• A safety monitoring committee — a small group, short of a formally chartered DSMB, that some sponsors use as an intermediate structure for moderate-risk trials.

Whichever arrangement a DSMP specifies, the same core elements apply: a named reviewer, a defined review schedule, a defined adverse-event reporting mechanism, and defined stopping criteria — documented in writing before the trial opens, not assembled retroactively.

References

• NIH, “NIH Policy for Data and Safety Monitoring,” NOT-98-084 (June 10, 1998), grants.nih.gov — requires a DSMP for every NIH-funded clinical trial, scaled to size, complexity, and risk, and generally requires a DSMB for multi-site trials with substantial participant risk.
• National Library of Medicine, “Requirements for Data Safety and Monitoring Plans,” nlm.nih.gov/ep/dsm.html — summarizes the required elements: monitoring progress and safety, an adverse-event reporting mechanism to the IRB/FDA/NIH, and data-accuracy/protocol-compliance assurance; notes that close monitoring by the study investigator may be adequate for small, low-risk trials.
• NIH institute-specific DSMP guidance, e.g. NIAID’s DSMP tutorial and decision tree, NIMH’s DSMP writing guidance, NIAAA’s and NIAMS’s data and safety monitoring guidelines — each institute publishes its own worked guidance applying the same underlying NIH policy.
• FDA, “Establishment and Operation of Clinical Trial Data Monitoring Committees” (Guidance for Clinical Trial Sponsors, March 2006) — recommends, without mandating, an independent data monitoring committee for industry-sponsored trials with elevated safety concern.

Related CASRAI content

• Data Safety Monitoring Board (DSMB) — the independent board a DSMP can (but does not have to) designate as its monitoring mechanism.
• Clinical Trial Phases — Phase III status is the strongest single trigger for requiring a full DSMB rather than PI-level monitoring.
• Pharmacovigilance in Clinical Research — the adverse-event and SAE reporting infrastructure a DSMP’s reporting mechanism plugs into.
• IRB (Institutional Review Board) — the ethics-review body a DSMP’s adverse-event reports also flow to, distinct from and in addition to safety-data monitoring.