The first priority of STAR*D is to complete and report the findings of the multisite, NIMH-funded study designed to determine prospectively the comparative effectiveness of different treatment options for patients with MDD. STAR*D will evaluate the comparative effectiveness of these treatments when they are used to either augment the previous treatment or as new treatments for depressions that do not respond satisfactorily to an initial selective serotonin reuptake inhibitor (SSRI) medication. Assignments to treatment options are randomized at all Levels of treatment (2, 2A, 3, and 4) following initial treatment with an SSRI (Level 1). Clinical outcomes include symptoms, functional status, side effect burden, quality of life, and patient satisfaction. These outcomes are evaluated by independent assessments masked to treatment assignments.

Four types of results are expected from the STAR*D effort: (1) Primary hypotheses - the main hypotheses of the study as defined in the STAR*D protocol (e.g., the comparison of clinical outcomes between different randomly-assigned treatments). (2) Secondary hypotheses - as stated in the STAR*D protocol (e.g., predictors of response vs. nonresponse within and among randomized treatments). (3) Secondary analyses - developed by STAR*D investigators and conducted on the STAR*D dataset which were not part of the original STAR*D protocol (i.e., not preplanned primary or secondary hypotheses). (4) Finally, the STAR*D dataset can be used to inform ancillary studies and, thereby, become part of ancillary study analyses.

The STAR*D Communications Committee has the primary responsibility of reviewing and approving all written publications and oral presentations. STAR*D also encourages the expeditious conduct of scientifically meritorious secondary analyses using the unique database that will become available. The STAR*D Communications Committee has the primary responsibility for the initial review and approval/disapproval of any secondary data analyses and analyses of the primary and secondary hypotheses detailed in the final STAR*D protocol.

The Director and Associate Director at each Regional Center, as well as Clinical Site investigators, are encouraged to develop proposals for secondary analyses, as well as to develop ancillary studies. Such analyses using the STAR*D database should be developed through a process outlined in these guidelines.

Data Analyses and the Release of Results: To maintain STAR*D scientific integrity, no analyses or release of STAR*D data are permitted without the STAR*D Communications Committee and STAR*D Executive Committee approval. Premature analyses or release of the primary endpoints of the trial cannot occur before release of main study results.

No Regional Center or Clinical Site may analyze data for internal or external use or for comparison or for incorporation with any other databases in which patients from STAR*D could be identified or derived. Investigators are not allowed to present data gathered from their Regional Center or Clinical Sites before the primary analyses are published. Results derived from STAR*D data may not be discussed with the news media without authorization of the STAR*D Communications Committee and the NIMH Project Officer (Dr. Niederehe).

Procedural Guidelines: The procedures for all types of analyses are given in detail in the following paragraphs and concern all form of publications, including manuscripts, abstracts, and posters. These procedures are intended to protect (1) the integrity of STAR*D, (2) the interests of all participants and investigators in the trial, (3) the scientific integrity of all study reports (i.e., to ensure they conform to the requirements of rigorous study design, are accurately presented, that authorship is appropriate, and that the text of all publications is clear and well-written). Authors who participate in the writing of manuscripts from STAR*D data (primary hypotheses, secondary hypotheses, secondary analyses, and ancillary study analyses) will do so in accordance with the International Committee of Medical Journal Editors guidelines (New Engl J Med 1991; 324:424-428).

First authors have the primary responsibility for organizing the writing group, directing the analyses, and drafting the manuscript. First authors are expected to include only those names in the final list of authors who have participated in the writing and/or analyses of the paper in accordance with those guidelines. Conflicts about authorship should be resolved, if at all possible, by the Communications Committee. Disputes that are not settled by the Communications Committee are referred to the Executive Committee, which makes the final decision. The order of co-authorship should be determined by the first author, according to the principles described below.

Release of Data: It is the policy of STAR*D to prohibit any release of data to anyone other than the Data and Safety Monitoring Board prior to the completion of the randomized trial. Other than endpoint studies, analyses may be conducted at any time and are encouraged so that STAR*D contributions to the field are furthered and STAR*D gains greater visibility in the scientific community by contribution of high-quality research. Oral presentations of the study rationale, objectives, and design are also encouraged to improve the visibility of the study. Such efforts will also provide an intellectual challenge for those participating in the study before the final major endpoint studies are published.

Primary and Secondary Hypotheses: STAR*D primary reports include reports of all primary and secondary hypotheses as stated in the written protocol. Primary reports include manuscripts from STAR*D that are solely concerned with the study design; the description of the STAR*D population at baseline; the evaluation of the treatment effects on the primary and secondary endpoints for Levels 1, 2, 2A, 3, and 4; the assessment of predictors of response for Levels 1, 2, 2A, 3, and 4 (secondary hypotheses); and the assessment of predictors of course during follow-up (secondary hypotheses).

All primary reports will include at least the principal and co-principal investigators (AJR and MF), at least two additional authors from the Dallas National Coordinating Center (NCC), at least one author from each of the Special Function Regional Centers (RCs) (Boston, New York, and Pittsburgh), and at least one author from NIMH. Authors of the primary reports will also include one representative author from the NIMH, and Drs. Wisniewski and Lavori as biostatisticians. Authors will likely include at least one analyst from the Pittsburgh Data Coordinating Center (DCC).

The contributions of all investigators involved in STAR*D who are not named as co-authors of the primary reports will be acknowledged as "for the STAR*D Investigators Group," with a list in the appendix of all the investigators at each Regional Center.

Primary Hypotheses: The Director and Co-Director of the Dallas National Coordinating Center (Drs. Rush and Trivedi) will be the primary authors for all manuscripts that report the results of the treatment effects on the primary and secondary clinical endpoints for Level 2. Drs. Trivedi and/or Rush will be among the first four authors listed on all other primary/secondary outcome reports.

The Director and Associate Director of the Pittsburgh Regional Center with Special Functions (Drs. Thase and Howland) will be the primary authors for all manuscripts that report results of the treatment effects on the primary and secondary clinical endpoints for Level 2A.

The Director and Associate Directors of the Boston Special Function Regional Center (Drs. Fava, Nierenberg, and Alpert) will be the primary authors for all manuscripts that report the results of the treatment effects on the primary and secondary clinical endpoints for Level 3.

The Director and Associate Directors of the New York Regional Center with Special Functions (Drs. Quitkin, McGrath, Stewart, and Sackeim) will be the primary authors for all manuscripts that report the results of the treatment effects on the primary and secondary clinical endpoints for Level 4.

Dr. Michael Kashner will be the primary author for all the manuscripts that report the results of the utilization and cost data described within the protocol.

Secondary Hypotheses (Predictors of Treatment Effects): Dr. Trivedi will be the primary author on all reports involving predictors of outcome using the primary and secondary outcomes for Level 1 acute phase. Drs. Rush and Fava will be primary authors for predictors of outcome across treatments at Level 2. Drs. Friedman, Howland, and Thase will be the primary authors for predictors of outcome to cognitive therapy (switch or augment) for Level 2. Drs. Fava, Nierenberg, and Alpert will be primary authors on predictors of outcome for Level 3 treatments. Drs. McGrath, Quitkin, Stewart, and Sackeim will be primary authors on predictors of outcome for Level 4 (acute).

Secondary Hypotheses (Course of Illness): Drs. Trivedi and Nierenberg will be primary authors on reports describing the course of illness during the naturalistic follow-up using primary or secondary clinical outcomes (Dr. Trivedi), as well as reports of predictors of the course of illness during follow-up. These predictors include baseline variables (obtained at the start of Level 1) and predictors obtained during acute treatment at relevant levels (e.g., trajectories of response; comparative treatment sequence leading to entry into follow-up) (Dr. Nierenberg).

Secondary Analyses: Secondary analyses are those analyses other than primary and secondary analyses and other than ancillary study analyses. Secondary analyses provide manuscripts from STAR*D which are not primary reports (i.e., those on (a) study design; (b) description of the STAR*D population at baseline; (c) evaluation of the treatment effects on the primary and secondary endpoints for Levels 1, 2, 2A, 3, and 4; (d) assessment of predictors of response for Levels 1, 2, 2A, 3, and 4; and (e) description of course of illness in follow-up or predictors of course in follow-up. For example, secondary analyses may compare subgroups defined at baseline (e.g., a comparison of primary vs. specialty care patients, or a comparison of outcome in different ethnic groups). It is important to extract optimal scientific information from all such STAR*D secondary analyses.

The Communications Committee will actively encourage primary and coauthorship in secondary analyses manuscripts by all Regional Center Directors and Associate Directors, especially those RCs that are outside Dallas, Boston, New York, and Pittsburgh. Authors of secondary analyses are restricted to investigators at participating STAR*D RCs or Clinical Sites.

Proposals for secondary analyses must be specified and submitted in writing to the STAR*D Communications Committee. The Communications Committee will review the request within one month of receipt and have a written response to the investigator within six weeks of submission. Once the analyses are approved, the proponents of such analyses will submit a "data analysis request form" (see Appendix 1) to the Pittsburgh DCC, which will conduct these analyses at no additional cost to STAR*D.

Secondary analyses reports will involve co-authorship by at least two investigators at the National Coordinating Center, one at the Boston Regional Center, and either Dr. Lavori or Wisniewski. The contributions of all investigators involved in STAR*D who will not be named as co-authors in reports of secondary analyses will be acknowledged as "for the STAR*D Investigators Group," with a list in appendix of all the Regional Centers and of investigators at each center. The manuscript needs to be submitted within two months from the receipt of the results of the analyses from the DCC.

When secondary analyses are proposed by any investigator, participation by other STAR*D investigators will be encouraged by the STAR*D Communications Committee. The STAR*D Communications Committee will circulate via e-mail the proposal to all Regional Center Directors and Associate Directors who, in turn, should circulate that information to relevant Clinical Site Directors, the Clinical Research Coordinators, fellows, or other interested parties. A response to the Communications Committee is expected within seven days by those interested in participating as coauthors (who may come from any Clinical Site or Regional Center). If too many express such an interest, the STAR*D Communications Committee will arbitrate the final listing with an eye toward increasing the likelihood of authorship for RCs with robust subject entry or retention rates.

If several applications for similar secondary analyses are received by the STAR*D Communications Committee, the Chairmen (Drs. Fava and Sackeim) will ask the applicants to resolve their differences and resubmit a joint proposal. If irreconcilable differences exist between the applicants, or if the applicants are unwilling to cooperate, the STAR*D Executive Committee will grade the applications according to the following criteria: scientific merit, previous experience of the investigators, balance to assure that projects are spread among investigators of all Regional Centers and Clinical Sites, and the performance of the Regional Center associated with the proposal (when indicated). The final decision concerning which secondary analyses proceed will be made based on this grading system.

Ancillary Study Reports: Before being conducted, all ancillary studies are reviewed and approved by the STAR*D Ancillary Studies Committee and by the STAR*D Executive Committee. Applicants for ancillary studies must obtain external, adequate funding to cover the costs involved in implementing and analyzing the study results. The cost to investigators of data acquisition and analyses for ancillary studies will be proportionate to the required effort. Such costs must be covered by the ancillary study investigators. Ancillary studies involve data gathering beyond the data collection required and supported by the STAR*D protocol. A specific ancillary study typically will use parts of the STAR*D database.

All ancillary study reports must be approved by both the STAR*D Communications Committee and the STAR*D Executive Committee. The Pittsburgh DCC will carry out the planned analyses for all ancillary studies (unless a specific different management is justified and accepted by the STAR*D Ancillary Studies Committee and the STAR*D Executive Committee)within 2 months from the submission of the request for analyses.

Co-authors of publications based on ancillary study findings, as well as of abstracts reporting on these studies, will include at least one investigator at the National Coordinating Center, one at the Boston Regional Center, and one at the Pittsburgh Data Coordinating Center. The contributions of all investigators involved in STAR*D, who are not named as co-authors of these ancillary study reports, will be acknowledged as "for the STAR*D Investigators Group," with a list in Appendix 1 of all investigators at each participating Regional Center. The manuscript needs to be submitted within two months from the receipt of the results of the analyses from the DCC.

Review Process for Published Papers: All reports (analyses of primary and secondary hypotheses, secondary analyses, ancillary study analyses) (initial and subsequent drafts) in the form of manuscripts, abstracts, or posters using the STAR*D database will be submitted to the Chairman (Dr. Fava) of the STAR*D Communications Committee for circulation to the full Committee for review. Two principal reviewers, after review by members of the Communications Committee, will provide detailed written reviews to the authors. These reviews are intended to ensure that study reports are scientifically valid and meritorious, are accurately presented, that authorship is appropriate, and the text of all publications is clear and well-written. The author's final draft with a list of the answers to the reviewers' comments will then be submitted to the STAR*D Communications Committee for approval.

Once approved, the manuscript (or abstract/poster) will be presented to the STAR*D Executive Committee, which will provide final review and approval within two weeks. A written communication from the STAR*D Executive Committee Chair (Dr. Rush) will be sent to the principal author upon approval. The final, revised manuscript (or abstract/poster) is then submitted. The primary author must notify the Chairman of the STAR*D Communications Committee (Dr. Fava) of the outcome of the submission, and to forward results of written reviews from the journal to which the manuscript is submitted. When responses/revisions are made in response to reviews by the journal, these responses and the final resubmission need to be cleared again by the STAR*D Communications Committee before resubmission.

Abstracts and Oral Presentations: As mentioned earlier, abstracts/posters follow a stepwise sequence to obtain concept approval and final approval similar to that of manuscripts. However, abstracts will not be approved if (1) the time between submission of the request to the Communications Committee and the abstract submission date is less than four months or (2) if the Pittsburgh DCC is unable to complete the required analyses within the required two-month time period.

Abstracts which have highest priority are those that can readily become peer-reviewed, published articles. First authors of abstracts are expected to convert this material into a submitted publication (draft to Communications Committee within two months following abstract presentation). Once an abstract has been presented, either poster or oral, the primary author has two months to either request additional analyses or submit a draft manuscript.

The STAR*D Communications Committee retains the authority to (1) change first author if these deadlines are not met, (2) decline subsequent abstract proposals by proposed first authors who have not previously converted abstract presentations into published reports.

First authors of abstracts will need to: (a) submit the previously approved data analysis request form to the Pittsburgh DCC, which conducts the analyses within two months (unless otherwise specified), (b) generate an abstract within two weeks after receipt of the results of the analyses, (c) submit the abstract to the STAR*D Communications Committee, which will approve, approve with revision, or reject the abstract within two weeks after receipt of the abstract, and (d) modify the abstract (whenever necessary) according to the review of the STAR*D Communications Committee and resubmit it within two weeks after receiving the response from the Committee. This process may, therefore, take up from a minimum of two months to a maximum of four months to be completed. As outlined earlier, proposed authors and order must be approved by the STAR*D Communications Committee.

Oral presentations about STAR*D, whether or not using data derived from STAR*D, must be submitted monthly on the attached form (Appendix ___) to the Operations Manager at the NCC.

ADD FORMS:
(1) Secondary Analyses
(2) Monthly Report of STAR*D Presentations

Data Request Form (Word)
Data Request Form (PDF)
Star*D Presentations July 2000 - May 2001 (PDF)