Evaluation of Outcomes with Citalopram for Depression Using Measurement-Based Care in STAR*D: Implications for Clinical Practice

• p. 9 Table 4 (continued), Agoraphobia, No (N=2,508), HAM-D columns

  728 29.1

Only the authors can say what is right here.

Judge's note: Page 9 image checked. Table 4 footnote a says subjects with missing exit HAM-D were classified as not in remission and that percentages are based on the number of subjects with each baseline characteristic. The HAM-D denominator should therefore be the row N. The neighbouring HAM-D cells all reproduce from the row N: GAD 641/2170 = 29.5, OCD 706/2437 = 29.0, panic 719/2472 = 29.1, agoraphobia Yes 55/335 = 16.4. This cell does not. The gap is 0.1 point, so it is trivial, and only the authors can say whether the count or the percentage is off.

Second check: another model agreed

• p. 6 Figure 3 caption and bars

  FIGURE 3. Percent of 2,876 Outpatients With Nonpsychotic Major Depressive Disorder Who Achieved Response or Remission

• p. 6 Figure 3 title

  Percent of 2,876 Outpatients With Nonpsychotic Major Depressive Disorder Who Achieved Response or Re-mission

Judge's note: Bar heights read from the p6 image: remission ≈11.4, 19.4, 22.1, 14.5, 12.9, 10.0, 12.9 (sum ≈103); response ≈17.8, 24.5, 22.9, 11.5, 11.0, 4.9, 7.6 (sum ≈100). The Results text describes the figure as the distribution of time to first remission/response among those who achieved it. If 2,876 were the denominator, the series would sum to only about 33% and 47%. The title mislabels the population.

Second check: another model agreed

• p. 7 Table 3, serious adverse events

  Any psychiatric serious adverse eventd 55 2.6 2 0.3 57 2.0

• p. 7 Table 3, Total

  Total 103 4.9 13 1.6 116 4.0

• p. 7 Table 3, serious adverse events

  Psychiatric hospitalization 50 — 2 — 52

Only the authors can say what is right here.

Judge's note: I looked at the Table 3 crop and page 7. The arithmetic part does not hold up as an error. The Results text on p.7 says "116 participants experienced at least one serious adverse effect; most of these patients (88.8% [N=103])...", so Total counts participants, and one participant can fall into several categories. A subcategory sum above the total (110 vs 103; 123 vs 116) is therefore expected. 'Any psychiatric' = 55, which is at most 50+6 = 56, so it is a valid union with one participant having both a psychiatric hospitalization and suicidal ideation. The remission column sums exactly (13). What remains is the footnote placement. The content of footnote c describes the 'Any psychiatric' row, and the content of footnote d describes the 'Psychiatric hospitalization' row, but the markers sit on other rows. Only the authors can confirm the intended placement.

Second check: another model agreed

• p. 1 Abstract, Results

  (especially anxi-ety disorders or drug abuse)

• p. 9 Table 4, Drug abuse/dependence

  <0.06

Only the authors can say what is right here.

Judge's note: On the p9 image, Table 4's drug abuse/dependence row shows OR 0.71 with p <0.06 for HAM-D and OR 0.75 with p <0.10 for QIDS-SR. The stated threshold is p<0.05 (p5). Alcohol abuse (<0.04) is the substance-use variable that meets the threshold. The number of comorbid disorders is significant (0.0001), so the main clause holds; only the drug-abuse example is unsupported.

Second check: another model agreed

• p. 2 Introduction, domain definitions

  2) social features (e.g., education, employ-ment status, income, insurance, and marital status)

• p. 11 Table 5, Demographic domain

  Education (reference group=high school but < college)

Judge's note: Checked the domain definitions on p2, Table 1 on p4 (text layer) and Table 5 on p11. The stepwise models were run within domains, so which domain education was entered into is left unclear. This is a minor consistency point.

Second check: another model agreed

• p. 9 Discussion, right column

  controlling for both severity and treatment settings

• p. 5 Statistical Analysis

  independent of the effect of regional center and baseline depression severity

Only the authors can say what is right here.

Judge's note: I compared the p.9 Discussion with the p.5 Statistical Analysis section ("independent of the effect of regional center and baseline depression severity") and with footnotes b and c of Table 4 (adjusted for regional center and baseline HAM-D or QIDS-SR). In Table 4, 'Setting' (primary vs psychiatric care) is a predictor, not a covariate. 'Treatment settings' therefore misdescribes the adjustment, unless it is meant loosely for regional centers. This is a wording issue.

Second check: another model agreed

• p. 6 Results, Pretreatment Correlates of Remission

  except that anxious depression and concurrent gener-alized anxiety disorder were also associated with lower QIDS-SR remission rates

• p. 9 Table 4, Generalized anxiety disorder, HAM-D p

  <0.05

• p. 8 Table 4, Gender / Marital status p values

  <0.002

• p. 6 Pretreatment Correlates of Remission

  anxious depression and concurrent gener-

Only the authors can say what is right here.

Judge's note: Table 4 on the p9 image: GAD shows HAM-D OR 0.80, p <0.05 and QIDS-SR OR 0.80, p 0.03. Anxious depression is the only listed factor that is significant for QIDS-SR alone (0.67 vs <0.002). Other factors differ the other way: gender is <0.002 vs <0.10 and marital status <0.02 vs <0.23 (p8), and alcohol abuse is <0.04 vs <0.09 (p9). Naming GAD as a QIDS-SR-only finding does not match the table under the stated p<0.05 threshold.

Second check: another model agreed

• p. 7 Results, final paragraph (Table 5 description)

  non-Caucasian, male, and less educated

• p. 11 Table 5, Race, Other

  1.11

• p. 8 Table 4, Race, Other (N=190)

  31.1

• p. 9 Discussion, left column

  Caucasian, female, better educated, and more highly paid

Judge's note: Checked Table 5 (p11): African American 0.69/0.66, Other 1.11/1.04. Table 4 (p8): Other 59 (31.1%), OR 1.14. Footnote d (p10 text) limits the significant contrast to Caucasian vs African American. The wording generalises that contrast to all non-Caucasians.

Second check: another model agreed

• p. 10 Discussion, left column

  married or cohabiting patients met criteria for treat-ment response with greater frequency than single par-ticipants

• p. 8 Table 4, Marital status

  Never married (N=823)

• p. 10 Discussion

  met criteria for treatment response with greater frequency than single par-

• p. 8 Table 4, Marital status

  Marital statuse

Only the authors can say what is right here.

Judge's note: I checked Table 4 on p.8 and the Discussion on p.10. Table 4 gives remission by marital status: HAM-D married 31.2% vs never married 24.7%, p<0.02; QIDS-SR p<0.23. The categories are never married, married, divorced and widowed. The paper defines response separately (≥50% QIDS-SR reduction, Figure 3), and no response or attrition results by marital status or living arrangement appear. The preceding sentence says 'remission', so 'response' is probably loose wording. Even so, the cohabitation and dropout statements are not backed by the data presented.

Second check: another model agreed

• p. 10 Table 4, footnote g

  < college significant different from ≥ college

Judge's note: The page 10 image shows the word as printed. Every parallel phrase in footnotes d–i reads "significantly different".

Second check: none (judge only)