Duration of untreated illness and antidepressant fluvoxamine response in major depressive disorder


Akiko Okuda, MD, Department of Psychiatry, Fujita Health University School of Medicine, Toyoake, Aichi 470-1192, Japan. Email: akkon-o@fujita-hu.ac.jp


Aims:  The aim of this study was to analyze the relation between treatment response and the duration of untreated illness (DUI) in 133 outpatients with the first major depressive disorder (MDD) episode.

Methods:  A logistic regression was performed with DUI, sex, age at onset, and score for 17 items on the Hamilton Depression Rating Scale at the time of start of fluvoxamine treatment as the explanatory variables, and the response and the remission as the outcome variables.

Results:  Regression analysis showed significant association between the response and DUI (P < 0.0001), and between the remission and DUI (P < 0.0001), respectively. The remission rate gradually decreased with longer DUI.

Conclusion:  Early treatment of first depressive episodes is important because a shorter DUI implied better remission outcomes.

TREATMENT OUTCOME OF major depressive disorder (MDD) has been shown to be poorer than expected in many cases.1 Therefore, both identifying factors that affect prognosis and investigating methods to cope with them are important in order to improve the effectiveness of treatment for MDD. Factors affecting prognosis that have been investigated to date are appropriateness of treatment,2 duration of untreated illness (DUI),3–6 sex,3,4,7 age at onset,8 severity of depression,3,9 social support,4,10 personality structure2,3,11 and stressful life events.3,10,12 However, the results of different studies have not always agreed, and many factors were not subject to intervention.

In this study, among clinical factors associated with MDD, we concentrated on DUI, which is highly amenable to intervention and is considered to be linked to improved treatment outcome.

Among reports to date on the prognosis and DUI of MDD, a six-month follow-up of 121 patients with MDD in a National Institute of Mental Health collaborative study13 showed that DUI is the most important factor in predicting treatment outcome. Hirschfeld et al.,14 Ionescu et al.3 and Javier de Diego-Adelino et al.6 reported that DUI is a predictive factor for treatment outcome. Patten5 showed that after the diagnostic criteria of MDD were satisfied, about 20% of patients recovered in the first week; however, after symptoms had continued for 6 months, the probability of recovery in the subsequent week was less than 1%.

In the present study, we examined the difference in treatment outcome with DUI in the first major depressive disorder (MDD) episode.



The subjects were patients older than 20 years old and younger than 75 years old. They were examined for first depressive episodes in the Department of Psychiatry, Fujita Health University Hospital in the period from July 1997 to August 2008. All subjects were diagnosed with MDD, based on DSM-IV, by a psychiatrist with appropriate clinical experience. All patients gave informed consent at the first visit to the hospital. In this study, there were no patients with recurrent depressive episodes. There was no antidepressant treatment until the visit to the hospital. All cases were assessed for treatment outcome after 8 weeks of treatment. The exclusion criteria for registration of the subjects were: (i) a history of schizophrenia or schizoaffective disorders; and (ii) a history of mania or mixed episodes. Patient information was used anonymously with care so that individuals could not be identified, and the study was approved by the Ethics Committee of Fujita Health University Hospital.

Clinical evaluation

At the start of treatment, the patients and their families were interviewed concerning the period from appearance of depressive mood until the time of treatment at the hospital (duration of untreated illness: DUI), and age at the time of this study. Onset of symptoms was defined as the time when the depressed mood developed. It was not the time when their condition satisfied five of the nine items for an MDD diagnosis based on DSM-IV. We asked the following fixed question, ‘When did the depressed mood develop?’ to the patients themselves or to their families during the first visit. DUI was determined from the onset of symptoms to the time of treatment. The patient's sex was also recorded. The severity of MDD at start of treatment (HAM0) was also evaluated with a 17-point score using the Hamilton Depression Rating Scale (HAM-D). A psychiatrist or a psychologist who had underwent training in the use of HAM-D conducted the HAM-D interview. The main therapeutic agent was fluvoxamine (FLV). The loading dose of FLV was 25 mg to 150 mg, at the discretion of the doctor. Only minor tranquilizers were permitted as concomitant medication. Concomitant use of major tranquilizers and psychotherapy such as cognitive behavioral therapy and interpersonal psychotherapy was prohibited during the study. We then observed the time course of the FLV response over eight weeks.

The treatment outcome was assessed at 0 weeks (HAM0) and at 8 weeks (HAM8) by a psychiatrist or a psychologist. The treatment outcome was evaluated based on both the response and the remission.15 Response was defined as a more than a 50% decrease at HAM8 from HAM0 [(HAM0–HAM8)/HAM0 ≥ 0.5]. Remission was defined as 7 points or less at HAM8 [HAM8 ≤ 7].

Data analysis

The following analyses were performed: (i) Analysis of basic patient characteristics. (ii) Analysis of the relation between treatment outcome and DUI. The explanatory variables were DUI, sex, age and HAM0. The outcome variables were both the response and the remission, which were used in logistic regression analysis. (iii) Additional analysis of remission rate: we classified DUI in five blocks and examined the remission rate in each for the continuity of the therapeutic results. The DUI in each block (A–E) were as follows. Blocks A–E: block A (24 patients): less than one month (2–3 weeks); block B (23 patients): more than one month and less than two months (4–7 weeks); block C (28 patients): more than two months and less than three months (8–11 weeks); block D (29 patients): more than three months and equal to or less than six months (12–24 weeks); and block E (29 patients): more than six months (≥25 weeks). Logistics regression analysis was performed with the response and the remission rate as the outcome variable and the five DUI blocks (A–E) as explanatory variables.


Patient characteristics

In this study, 679 patients diagnosed with MDD were examined in the Department of Psychiatry of Fujita Health University Hospital. Among these patients, 231 patients were diagnosed with a first episode of MDD. As far as we could determine in their therapeutic history, they received no antidepressant treatment before visiting the hospital. Informed consent was received from all participants in this study. Ninety-eight patients dropped out (42%) and their treatment outcome could not be evaluated at eight weeks. As a result, the analysis was conducted with 133 patients for whom complete evaluation of treatment outcome was possible. We were unable to get sufficient information from lost samples.

The mean DUI was 19.4 ± 22.4 weeks. There were 71 men and 62 women in the patient group, whose mean age at study entry was 44.1 ± 15.7 years. HAM0 was 20.8 ± 6.4 points and HAM8 was 9.8 ± 6.9 points. FLV dose was 25–150 mg/day, as prescribed by the physician. The mean dose ± SD was 50.8 ± 20.8 mg/day at the start of treatment and 117.7 ± 40.5 mg/day at week 8. The number of patients who responded to the treatment was 75 (56.4%), and remission occurred in 58 patients (43.6%).

Analysis of treatment outcome and DUI

The results of logistic regression analysis of DUI and sex, age and HAM0, with the response as the outcome variable, are shown in Table 1. A significant association was found between the response and DUI (P < 0.0001, odds ratio [OR], 1.07; 95% confidence interval [CI], 1.04–1.10).

Table 1.  Results of logistic regression analysis with response and remission
  1. CI, confidence interval; DUI, duration of untreated illness; HAM0, severity of major depressive disorder at start of treatment according to the Hamilton Depression Rating Scale; OR, odds ratio.

Study age0.220.13
DUI (weeks)0.0011.061.02–1.090.0011.101.03–1.13

The results of a logistic regression analysis of remission and DUI, sex, study age and HAM0 are also shown (Table 1). A significant association was found between the remission and DUI (P < 0.0001, OR, 1.10, 95%CI = 1.05–1.15). However, no association was found between treatment outcome and the other explanatory variables.

Further analysis of treatment outcome and DUI

Next, we examined the time until patients received initial treatment.

  • 1Patient data are shown in Table 2 with the DUI classified into five blocks. The mean HAM0 score ± SD in each block was 21.6 ± 6.5 in A, 20.4 ± 6.2 in B, 21.2 ± 6.7 in C, 20.3 ± 5.6 in D and 20.3 ± 7.1 in E. We show the response rate and the remission rate in Figures 1 and 2. The percentages achieving remission in each block were 70.8 in A, 65.2 in B, 57.1 in C, 24.1 in D and 10.3 in E. Thus, the remission rate was gradually lower as DUI became longer (Fig. 2).
  • 2Logistic regression analysis in the five DUI blocks, A, B, C, D and E, with the remission as the outcome variable, showed significant associations in B, C, D, and E (Table 3).
Table 2.  Demographic and clinical characteristics for five blocks (A to E)
  1. FLV, fluvoxamine; HAM0, severity of major depressive disorder at start of treatment according to the Hamilton Depression Rating Scale; HAM8, treatment outcome at 8 weeks according to the Hamilton Depression Rating Scale.

No. of weeks to treatment2–3 weeks4–7 weeks8–11 weeks12–24 weeks25 weeks
No. of patients2423282929
Gender (male : female)14:105:1816:1221:815:14
Study age (mean age ± SD)40.2 ± 14.845.7 ± 15.542.9 ± 15.142.5 ± 12.946.6 ± 19.2
HAM0 (mean points ± SD)21.6 ± 6.520.4 ± 6.221.2 ± 6.720.3 ± 5.620.3 ± 7.1
HAM8 (mean points ± SD)6.4 ± 5.77.0 ± 5.68.5 ± 7.511.6 ± 6.414.4 ± 5.8
FLV dose at start of treatment(Mean mg ± SD)51.0 ± 13.850.0 ± 22.650.0 ± 19.250.4 ± 20.552.5 ± 26.5
No. of responses (%)18 (75.0)18 (78.2)20 (71.4)12 (41.4)7 (24.1)
No. of remissions (%)17 (70.8)15 (65.2)16 (57.1)7 (24.1)3 (10.3)
Figure 1.

ercentage achieving response in duration of untreated illness for five blocks: (inline image) A, 2–3 weeks; (inline image) B, 4–7 weeks; (inline image) C, 8–11 weeks; (inline image) D, 12–24 weeks; (inline image) E, ≥25 weeks.

Figure 2.

Percentage achieving remission in duration of untreated illness for five blocks: (inline image) A, 2–3 weeks; (inline image) B, 4–7 weeks; (inline image) C, 8–11 weeks; (inline image) D, 12–24 weeks; (inline image) E, ≥25 weeks.

Table 3.  Logistic regression analysis of DUI (B to E) with the remission rate
DUIP-valueOdds ratio95%CI
  1. CI, confidence interval; DUI, duration of untreated illness.



We examined the treatment outcome in 133 patients with a first MDD episode in response to 8 weeks of treatment mainly with FLV. The outcome variables were response and remission, and the explanatory variables were DUI, sex, study age and HAM0. The results showed that DUI was related to both the response and remission. However we did not find any association between treatment outcome and the other explanatory variables. Only DUI showed a significant association in both the response and the remission. Among the explanatory variables that we investigated, DUI was considered to be the most important factor influencing treatment outcome in patients with MDD.

It is important to know when to start treatment in order to achieve a good treatment outcome and a high rate of remission. We show the response and remission rates in MDD in Figures 1 and 2. The response rate was higher in group B than A, but it gradually decreased afterwards. The remission rate decreased gradually as the DUI became longer. In order to obtain good quality of life in MDD, long-lasting remission is thought to be important.16 We think that remission is more strongly associated with MDD outcome than response. We therefore focused more on the remission rate than the response rate. When the relation with remission was examined by dividing the DUI into five blocks, the percentage with remission was found to be 10.3–70.8%. Theremission rate was gradually lower as DUI became longer (Fig. 2).

We compared our results with the results of a simulation analysis by Patten.5 When treatment was started, treatment outcome became higher with shorter DUI. The subsequent recovery rate dropped rapidly. In our study using standard FLV treatment, the remission percentage was 43.6%. The remission rate (43.6%) in our study was considerably higher than in the study by Patten (20%). The reason why the treatment outcome was higher in our group is that all patients in our study were treated with FLV, while there were treated and untreated groups in the study by Patten.5 Another study, in contrast, reported that treatment outcome is not affected by dosage of FLV17.

Therefore, the dosage of FLV at the discretion of the doctor was not reviewed in this study.

The present data contain insufficient information on dropouts, and this might have had some effect on the results. Further research is necessary regarding other factors that might be related to prognosis, such as the personalities of the patients themselves, family history, patients' tastes and appearance of adverse reactions. Limitations in this study include the fact that the DUI was a retrospective parameter for both the patients themselves and their families, and the period could not be determined exactly.

If treated with FLV, the percentage achieving the remission can be maintained at 43.6% and the results of FLV treatment are better than the results of previously reported data including untreated groups.5 Our calculations of the remission rate in each block of DUI showed that the remission rate decreased as the DUI became longer.

In conclusion, we suggest that early treatment of first depressive episodes is important as a shorter DUI was associated with better remission outcomes.


We thank Ms R Nobata, Ms K Kariya, Ms H Ieda, Ms R Ando, Ms T Kitajima, Mr K Magnuson and Mr T Furukawa for their technical support. This work was supported in part by research grants from the Ministry of Education, Culture, Sports, Science and Technology, the Ministry of Health, Labor and Welfare, and the Japan Health Sciences Foundation (Research on Health Sciences focusing on Drug Innovation).