|Year : 2020 | Volume
| Issue : 3 | Page : 217-224
Evaluation of Family Psychoeducation Efficacy in Schizophrenia: A Way to Do it More Simply for Routinely Monitoring
Yann Hode1, Aline Deruyver2, Regine Vonthron3, Claudine Clement4, Wydad Hikmat5, Said Fattah6
1 SAMSAH/SAVS ARSEA 1 Faubourg des Vosges 68920 Wintzenheim, France
2 Université de Strasbourg Laboratoire ICube, UMR 7357 Equipe CSTB - 300 bd Sébastien Brant - CS 10413 - F-67412 Illkirch Cedex, France
3 Hôpitaux Civils de Colmar, 9 Avenue de la Liberté, 68024 Colmar, France
4 Centre Hospitalier VALVERT, 78 Boulevard des Libérateurs 13391 Marseille cedex 11, Mulhouse, France
5 Hôpital Psychiatrique Kelaâ des Sraghna, MAROC, Maroc
6 WASP Section on Family Intervention Programs, Cabinet de Psychiatrie le Trident 36 rue Paul Cézanne, 68200, Mulhouse, France
|Date of Submission||03-Oct-2020|
|Date of Decision||01-Dec-2020|
|Date of Acceptance||03-Dec-2020|
|Date of Web Publication||24-Dec-2020|
Dr. Yann Hode
6 Rue Lazare Weiller, 67600 Sélestat
Source of Support: None, Conflict of Interest: None
Background: Despite many control trials and meta-analyses proving the positive effect and the economic interest of family psychoeducation in schizophrenia, this practice remains marginal. There may be several reasons behind the clinicians' lack of motivation to implement such programs. Simple and common evaluation tools to compare program efficacy, improve the programs' content, and guaranty the fidelity of the delivery over time by routinely monitoring the efficacy level would be a way (among many) to stimulate clinical interest in psychoeducation. Two common self-administered questionnaires used in combination, the Center for Epidemiologic Studies Depression (CESd) to assess caregivers' depressive symptoms and 20-item Life Skill Profile (LSP-20) to assess patient functioning, would be good candidates for that. Methods: Family caregivers (n = 78) of seven consecutive groups participating in a program named Profamille filled in these questionnaires before and at the end of the program and 1 year later. More than the half of the caregivers have an initial depressive score indicating a higher risk of cardiovascular mortality/morbidity. Results: Both CESd and LSP20 significantly changed 1 year later. The size effect for CESd was 1.4 and for LSP20 was 0.6. These changes were linked to a statistically significant reduction in the number of hospitalized days for the patients (by a factor 3) and a number of days lost at work for the caregivers (by a factor 4). Conclusions: These two questionnaires are therefore useful to compare different programs, obtain preliminary evidence from exploratory trials, like for the Profamille V3 program, and help ensure the fidelity of program delivery.
Keywords: Center for Epidemiologic Studies Depression, family intervention, life skill profile, profamille, psychoeducation, schizophrenia
|How to cite this article:|
Hode Y, Deruyver A, Vonthron R, Clement C, Hikmat W, Fattah S. Evaluation of Family Psychoeducation Efficacy in Schizophrenia: A Way to Do it More Simply for Routinely Monitoring. World Soc Psychiatry 2020;2:217-24
|How to cite this URL:|
Hode Y, Deruyver A, Vonthron R, Clement C, Hikmat W, Fattah S. Evaluation of Family Psychoeducation Efficacy in Schizophrenia: A Way to Do it More Simply for Routinely Monitoring. World Soc Psychiatry [serial online] 2020 [cited 2022 Aug 10];2:217-24. Available from: https://www.worldsocpsychiatry.org/text.asp?2020/2/3/217/304817
| Introduction|| |
Since the seminal 1980 articles by Anderson, Hogarty, and Reiss on family psychoeducation in schizophrenia, many psychoeducation methods have been validated to improve the lives of people with schizophrenia and their families. Moreover, the economic efficiency of psychoeducation has also been demonstrated., Despite many positive randomized controlled trials and meta-analyses,,,,, family psychoeducation has still not become widely available to family members of patients with schizophrenia.,,
Psychoeducation is as effective at preventing relapse as an antipsychotic treatment and does so at a lower cost. Why is the diffusion of psychoeducation not as widespread as the use of antipsychotic drugs? One of the most straightforward explanations is that giving a pill requires less effort than organizing a psychoeducational group.
The lack of motivation by clinicians to put an effort into psychoeducation may stem from an inability to perceive the benefits for two reasons. First, many outcome measures such as burden, quality of life, or relapse may be not marked, relevant, or familiar enough for the majority of clinicians. Second, many programs may have an effect not striking enough to be convincing. More research needs to be performed to improve program efficacy and make evaluation tools simpler to ensure reliable programs with proven efficacy.
Furthermore, the way a program is executed may quickly deviate from the way it was initially designed. This lack of reliability over time may also dissuade clinicians from putting efforts into psychoeducation. Maintaining the evaluation process initially used when a program was tested under controlled conditions is a way to prevent this deviation. However, a rigorous evaluation of how patients and families function may require meetings with trained professionals. This is costly and time consuming. Moreover, it is often difficult to meet patients because most patients have a poor understanding of their illness and may be reluctant to cooperate in an evaluation. Furthermore, to overcome barriers to implementing family psychoeducation, several practices have been developed: Either the patients are not included in the sessions, or the unit offering the program may not be involved in the patients' care. The result is that there is no direct access to patients to assess their functioning. Subsequently, there is a need for simple evaluation tools to overcome these constraints and help ensure more widespread application of psychoeducation programs.
What should the outcome measures be from a public health point of view? Patient well-being or functioning is interesting because this is what these programs intend to achieve. However, reducing costs for patients is also an important concern. Family member well-being is also often cited as an interesting outcome measure for several reasons. First, a suffering family will express a high level of emotion, which can contribute to patient relapse. Second, a suffering family may be too exhausted to help the patient, which would, in turn, affect the patient. Third, many studies have shown that people suffering from depression have a higher risk of mortality or morbidity. Suffering family members may become depressed, leading to further health consequences and health costs.
The number of workdays lost by caregivers and the number of hospitalization days of patients are good outcome measures, but they are rare events. They are on/off variables that do not enable the progressive effect on all participants to be assessed. Even if they constitute the main outcomes of interest, they would not be appropriate for preliminary studies with a small number of participants.
Self-administered questionnaires completed by participants only would be among the easiest tools to use for outcome evaluation. They do not require collaboration from patients or professional evaluation time or skills. However, such subjective evaluations may not reflect the reality of the health impact on the patient and family caregivers of psychoeducation.
This study sought to examine the efficacy of a family psychoeducation program on schizophrenia management using such tools. We want to know if these tools are sensitive to changes in our program and if they reflect an objective health improvement in patients and families.
| Methods|| |
This retrospective, observational study was conducted on seven consecutive groups of family members participating in a psychoeducational program called “Profamille version V3.” This program was carried out by a psychiatric team (psychiatrists and nurses) in France. Self-administered questionnaires were completed by participants at the beginning and at the end of the program. The sensitivity of these questionnaires and their relevancy to assess the program's effectiveness is studied. As this was an observational retrospective study conducted in accordance with French law, ethics committee approval and informed consent were not required.
Participants were recruited following a publication in a local newspaper or by referral from a family organization or a physician. The participants were all members of a family of a patient with schizophrenia or a related disorder. This program was offered to participants free of charge (the real cost was approximately the monthly salary of a nurse per participant).
All interested family members were invited to an information meeting about 3 months before the start of the program. Information about the aims, the rationale, and the content of the program as well as the session schedule were given during this meeting.
After the information meeting, and if they agreed to follow the entire program according to the given schedule, participants were enrolled in the order in which they appeared on the contact list. If they could not participate but were interested in a future group, their position on the contact list was maintained so that they could be invited to the next information meeting.
The psychoeducational program
Profamille is a psychoeducational program for families of patients with schizophrenia spectrum disorders. The patients are not invited to the program. The department offering the program generally does not know the patients. The third version (V3) of this program has 14 sessions of 4 h each, scheduled about once a week. These sessions are followed by maintenance sessions 3, 6, and 12 months later. Each group has about 10–12 participants. There is a program manual with very precise descriptions of what should be said or done according to different situations.
To assess patient functioning, the 20-item Life Skill Profile (LSP20) was chosen., The questionnaire was designed to allow caregivers and families to evaluate patient behavior without special training. Each item is scored on a scale of 1–4 where 4 is considered the lowest functional level. This questionnaire has been validated in several languages. It is used in many studies evaluating schizophrenia functioning,,,, for the purposes of assessing rehabilitation programs, for example. In our study, each item is scored according to patient functioning over the past 2 months.
Each participant is also asked about the number of days of patient hospitalization in the year preceding the first session of the program and the year following the last session.
To assess the level of emotion expressed by family members, which predicts patient relapse and the level of suffering of these family members, the 20-item, self-administered Center for Epidemiologic Studies Depression (CESd) Scale has several advantages: It measures the severity of depressive symptomatology during the past 7 days, has already been used in several studies on caregivers,,, reflects the level of expressed emotion, and predicts cognitive decline or morbidity in people with a high score. It has been validated in several languages and it has good acceptability. This questionnaire is completed about 3 months before the first session (waiting period), on the day of the first session, after the 14th (last) session, and 1 year after the last session. Since we can expect less morbidity if there are fewer or less intense depressive symptoms, at the first session and 1 year after the last session, each participant was asked about the number of days they missed work for illness in the last 12 months.
We compare CESd and LSP scores at the first and last sessions and 1 year later. The Friedman test was used. This is a nonparametric test equivalent to ANOVA that compares three or more paired groups. If significant, the Dunn's multiple comparison test was performed. P < 0.05 was considered significant in all analyses. The percentage of workdays lost by participants and patient hospitalization days in the year prior to the first session and the year following the last session were compared using the Chi-square test.
Since a CESd score of over 16 indicates greater depressive symptoms and a higher risk of morbidity/mortality,,, we separated people below this score from those over this score because pooling together depressed and nondepressed participants to observe mood improvement is not relevant.
The number of participant workdays lost may be linked to their depressive state and the number of patient hospitalization days may be linked to their functioning. For this reason, we check if LSP or CESd improvement at the last session is followed by a reduction in the number of workdays lost or hospitalization days 1 year later. Considering the rarity and the “all-or-nothing” nature of lost workday and hospitalization day events together with the complexity of the factors impacting on these events, any correlation analysis has little pertinence and odd-ratio (OR) analysis on subgroups is more appropriate.
Missing data analysis
Data missing prior to the completion of the 14 sessions or at the session following the 14th session might be due to a decline in the family member's mood or patient functioning growing worse, creating a sense of hopelessness. These two situations could negatively affect the motivation to attend the 1-year evaluation meeting. The previous CESd and LSP scores are shown to analyze this hypothesis.
It is not possible to infer the value of missing data, but in the worst-case scenario, one can hypothesize that all missing data correspond to a decline in CESd and LSP scores. To simulate this worst-case scenario, we used the worst previous score to replace missing scores and added an estimate of the mean score degradation. This estimate was calculated by averaging the paired difference between worsening scores. Then, initial scores and scores after 1 year were compared using the Wilcoxon signed-rank test for paired data.
All statistical calculations were performed using R software.
| Results|| |
The flowchart of the study is presented in [Figure 1] and sample characteristics in [Table 1]. For people whose data were incomplete, we tried to call them or obtain information about their mood change and global patient functioning through other members of their psychoeducational group (people of the same group often stay in touch).
|Table 1: Descriptive statistics of participants and their close relatives suffering from schizophrenia-spectrum disorder|
Click here to view
Effects of the program
Of the 72 participants who completed the 14 sessions, 51% had an initial CESd score above 16 and 60% of them had a score that had normalized by the last session (two-tailed Wilcoxon signed-rank test on paired data P < 0.0001, size effect = 1.4). Participants have the same initial CESd score as the CESd score 3 months before (waiting list). No CESd change is observed in the group of participants with an initial nondepressed CESd score [Figure 2] and [Table 2]. The CESd and LSP scores for those whose patient was evaluated 1 year later are presented in [Table 2].
|Figure 2: Mean Center for Epidemiologic Studies Depression score (standard deviation) at different times (waiting list, initial, final, 1 year) in initially depressed (black bar) and not depressed caregivers (white bar)|
Click here to view
|Table 2: Center for Epidemiologic Studies-Depression and Life Skill Profile 20 score changes at different stages of the program|
Click here to view
The CESd and LSP scores were lower 1 year after the last session. Participants whose LSP evaluations were not complete after 1 year were slightly more depressed and slightly younger, and their patients were slightly less functional [Table 3], although these differences were not statistically significant. However, these participants reported fewer patient hospitalization days (1.64%, Chi-square P < 0.0001). Among these participants, the nonretired were younger than nonretired participants who completed the 1-year evaluation (43 [9.2] vs. 53 [4.9] [Mann–Whitney U-test P = 0.0019]), their moods were similar (CESd; mean [standard deviation]: 12 [6.6] vs. 13 [7.9]), and they only missed 0.15% workdays (Chi-square P < 0.0001).
|Table 3: Characteristics of the participants lost to follow-up at 1 year|
Click here to view
Relationship between the score change at the 14th session and workdays lost and the number of hospitalization days 1 year later
In each case, we have two 2 × 2 table (factor A = group, factor B = number of days lost or hospitalized versus number of regular days), with two strata: 1 year before and 1 year after psychoeducation. To test if there is a difference between the two strata, we used the Breslow–Day test on the OR.
In the two cases, the Breslow–Day test on homogeneity of odds ratios gives a P < 2.2e-16. These results are in agreement with the hypotheses that the normalization of a depressive score predicts a greater reduction in the number of workdays lost and an improvement in the functioning score predicts a greater reduction in hospitalization days.
Worst-case simulation at 1 year
CESd change – each 1-year missing data value was completed with the highest (worst) of the two previous scores plus 8.2 (upper limit of the 95% confidence interval for the mean increase in scores that grew worse). With this correction, the CESd scores were lower at 1 year (two-tailed Wilcoxon signed-rank test P = 0.0496).
LSP change – Each 1-year missing data value was completed with the highest (worst) of the two previous scores plus 6.9 (upper limit of the 95% confidence interval for the mean increase in scores that got worse). With this correction, the LSP scores were lower at 1 year (two-tailed Wilcoxon signed rank test P = 0.026).
For participants whose evaluations were complete after 1 year, the observed pre-program percentages of workdays lost and hospitalization days for their patients were far below those of the other group. If we assumed that 1 year after these percentages corresponded with the percentages that the other group had before the program, we would obtain 118 missed workdays (instead of 5 before the program) and 665 hospitalization days (instead of 96 before the program). This drastic change may be seen as a worst-case simulation. By pooling these simulated data with those of the other group, we observe that the percentage of patient hospitalization days changes from 9.3% to 5.8% (P < 0.0001) and the percentage of missed workdays for caregivers changes from 1.8% to 0.9% (P < 0.0001).
| Discussion|| |
The CESd and LSP20 questionnaires were able to detect distinct positive effects of the Profamille V3 program and the changes observed in scores over time are relevant to the expected effects.
The CESd score improvement demonstrates that 65% of families with a score indicating a health risk had a “safe” score by the end of the 14th session (P < 0.01). Moreover, this result is stable 1 year later (size effect = 1.66). This decrease in morbidity and mortality risk is combined with a fourfold decrease in the number of workdays lost for illness during the year following the 14 sessions (P < 0.0001).
LSP20 scores mainly improved 1 year after the program (P < 0.001) and the size effect was 0.69. This delay reflects the time necessary to obtain a stable behavioral change. This improvement is confirmed by a threefold reduction in the number of patients' hospitalization days (P < 0.0001). The cost of the hospitalization days that were prevented by the program during the 1st year is approximately twice the cost of the psychoeducation program.
The LSP20 score observed and CESd score changes are linked to real changes impacting health and economics [patient hospitalization days or workdays lost by participants for illness, [Table 4]. These changes, which occurred over 1 year, were predicted by the score changes for the two questionnaires at the end of the 14th session [Figure 3]. This observation confirms the value of these two self-administered questionnaires as surrogate indicators to assess the efficacy of a psychoeducational intervention program.
|Table 4: Objective changes for participants (missed working days) and patients (number of days spent in hospital) before and after the program|
Click here to view
|Figure 3: (Left) Average change of the percentage of hospitalization days during the year if the Life Skill Profile scores are worse or better between the first and the final session with odd-ratio = 29.3 1 year before and 0.43 1 year after (right): Average change of the percentage of workdays lost if the participants are initially depressed (Center for Epidemiologic Studies Depression >16) and normalized their scores at the last session the other cases with odds- -ratio = 1.47 1 year before and odds ratio = 4.41 1 year later)|
Click here to view
Improved LSP scores could be interpreted as more positive participant thinking about the patients rather than a real improvement in patient functioning. Even if this hypothesis cannot be ruled out, we observed only a slight change in LSP scores after the 14th session, whereas we observed a great improvement in CESd scores. The LSP scores were greatly improved 1 year later, whereas the CESd scores were close to the previous ones [Table 2]. If the participant's mood improvement was responsible for a more positive perception of patient functioning, the LSP scores would immediately follow the CESd scores, which was not the case. Moreover, initial depressed caregivers whose CESd score was worse at 1 year reported a slight improvement of the LSP20 score of their relatives. Although this result is not significant because it concerns only three caregivers, it supports the hypothesis that LSP20 may be improved independently of the caregivers' mood.
The greatest limitation of our study was the missing data at 1 year. However, if missing data are completed with the worst values, LSP and CESd are still better after 1 year and the Profamille V3 program still improves the percentage of caregiver workdays lost and the percentage of patient hospitalization days.
| Conclusions|| |
This study showed that LSP20 and CESd questionnaires are good candidate measures to evaluate the effect of psychoeducation programs on patients and family caregivers, and the combined use of these questionnaires is relevant from public health and economic point of view. It also showed the Profamille V3 program's effectiveness on patients and caregivers in an observational, not randomized condition. CESd scores also provide information that may support family commitment and empowerment: Half of the families attending a psychoeducational program suffer in two ways (being at risk of early death and at risk of having a child with schizophrenia). This early death risk may be markedly reduced with an appropriate intervention that could also improve patient functioning. These two questionnaires are useful to compare different programs, obtain preliminary evidence from exploratory trials, like for the Profamille V3 program, and help ensure the fidelity of program delivery. These questionnaires will be helpful to ensure preliminary homogenization of practices among the sixty teams across French-speaking countries (Belgium, France, Morocco, Switzerland, etc.) that offer the Profamille V3 program and will be the prelude to a multicentric study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Anderson CM, Hogarty GE, Reiss DJ. Family treatment of adult schizophrenic patients: A psycho-educational approach. Schizophr Bull 1980;6:490-505.
Lucksted A, McFarlane W, Downing D, Dixon L. Recent developments in family psychoeducation as an evidence-based practice. J Marital Fam Ther 2012;38:101-21.
Shimodera S, Furukawa TA, Mino Y, Shimazu K, Nishida A, Inoue S, et al.
Cost-effectiveness of family psychoeducation to prevent relapse in major depression: Results from a randomized controlled trial. BMC Psychiatry 2012;12:40.
Breitborde NJ, Woods SW, Srihari VH. Multifamily psychoeducation for first-episode psychosis: A cost-effectiveness analysis. Psychiatr Serv 2009;60:1477-83.
Sin J, Norman I. Psychoeducational interventions for family members of people with schizophrenia: A mixed-method systematic review. J Clin Psychiatry 2013;74:e1145-62.
Xia J, Merinder LB, Belgamwar MR. Psychoeducation for schizophrenia. Cochrane Database Syst Rev 2011:CD002831.
Lyman DR, Braude L, George P, Dougherty RH, Daniels AS, Ghose SS, et al.
Consumer and family psychoeducation: Assessing the evidence. Psychiatr Serv 2014;65:416-28.
Bäuml J, Pitschel-Walz G, Volz A, Engel RR, Kessling W. Psychoeducation in schizophrenia: 7-year follow-up concerning rehospitalization and days in hospital in the Munich Psychosis Information Project Study. J Clin Psychiatry 2007;68:854-61.
Magliano L, Fiorillo A, Malangone C, De Rosa C, Maj M. Patient functioning and family burden in a controlled, real-world trial of family psychoeducation for schizophrenia. Psychiatr Serv 2006;57:1784-91.
Rummel-Kluge C, Pitschel-Walz G, Bäuml J, Kissling W. Psychoeducation in schizophrenia – results of a survey of all psychiatric institutions in Germany, Austria, and Switzerland. Schizophr Bull 2006;32:765-75.
Rummel-Kluge C, Kluge M, Kissling W. Frequency and relevance of psychoeducation in psychiatric diagnoses: Results of two surveys five years apart in German-speaking European countries. BMC Psychiatry 2013;13:170.
Brent BK, Giuliano AJ. Psychotic-spectrum illness and family-based treatments: A case-based illustration of the underuse of family interventions. Harv Rev Psychiatry 2007;15:161-8.
Leucht S, Tardy M, Komossa K, Heres S, Kissling W, Salanti G, et al.
Antipsychotic drugs versus placebo for relapse prevention in schizophrenia: A systematic review and meta-analysis. Lancet 2012;379:2063-71.
Dixon L, McFarlane WR, Lefley H, Lucksted A, Cohen M, Falloon I, et al.
Evidence-based practices for services to families of people with psychiatric disabilities. Psychiatr Serv 2001;52:903-10.
Hooley JM. Expressed emotion and relapse of psychopathology. Annu Rev Clin Psychol 2007;3:329-52.
Cuijpers P, Vogelzangs N, Twisk J, Kleiboer A, Li J, Penninx BW, et al.
Comprehensive meta-analysis of excess mortality in depression in the general community versus patients with specific illnesses. Am J Psychiatry 2014;171:453-62.
Welch CA, Czerwinski D, Ghimire B, Bertsimas D. Depression and costs of health care. Psychosomatics 2009;50:392-401.
Rosen A, Trauer T, Hadzi-Pavlovic D, Parker G. Development of a brief form of the life skills profile: The LSP-20. Aust N Z J Psychiatry 2001;35:677-83.
Mohr S, Simon A, Favrod J, Fokianos C, Ferrero F. Validation of the French version of the life skills profile with people suffering of schizophrenia. Encephale 2004;30:343-51.
Rosen A, Hadzi-Pavlovic D, Parker G. The life skills profile: A measure assessing function and disability in schizophrenia. Schizophr Bull 1989;15:325-37.
Parker G, Hadzi-Pavlovic D. The capacity of a measure of disability (the LSP) to predict hospital readmission in those with schizophrenia. Psychol Med 1995;25:157-63.
Kirkby K, Daniels B, Jones I, McInnes M. A survey of social outcome in schizophrenia in Tasmania. Aust N Z J Psychiatry 1997;31:405-10.
Favrod J, Grasset F, Spreng S, Grossenbacher B, Hodé Y. Benevolent voices are not so kind: The functional significance of auditory hallucinations. Psychopathology 2004;37:304-8.
Aki H, Tomotake M, Kaneda Y, Iga J, Kinouchi S, Shibuya-Tayoshi S, et al.
Subjective and objective quality of life, levels of life skills, and their clinical determinants in outpatients with schizophrenia. Psychiatry Res 2008;158:19-25.
Farreny A, Aguado J, Ochoa S, Haro JM, Usall J. The role of negative symptoms in the context of cognitive remediation for schizophrenia. Schizophr Res 2013;150:58-63.
Pickett-Schenk SA, Cook JA, Steigman P, Lippincott R, Bennett C, Grey DD, et al.
Psychological well-being and relationship outcomes in a randomized study of family-led education. Arch Gen Psychiatry 2006;63:1043-50.
Magaña SM, Ramírez García JI, Hernández MG, Cortez R. Psychological distress among Latino family caregivers of adults with schizophrenia: The roles of burden and stigma. Psychiatr Serv 2007;58:378-84.
Chiu MY, Wei GF, Lee S, Choovanichvong S, Wong FH. Empowering caregivers: Impact analysis of Family Link Education Programme (FLEP) in Hong Kong, Taipei and Bangkok. Int J Soc Psychiatry 2013;59:28-39.
Wolf JM, Miller GE, Chen E. Parent psychological states predict changes in inflammatory markers in children with asthma and healthy children. Brain Behav Immun 2008;22:433-41.
Raji MA, Reyes-Ortiz CA, Kuo YF, Markides KS, Ottenbacher KJ. Depressive symptoms and cognitive change in older Mexican Americans. J Geriatr Psychiatry Neurol 2007;20:145-52.
Brown JM, Stewart JC, Stump TE, Callahan CM. Risk of coronary heart disease events over 15 years among older adults with depressive symptoms. Am J Geriatr Psychiatry 2011;19:721-9.
Wilcox H, Field T, Prodromidis M, Scafidi F. Correlations between the BDI and CES-D in a sample of adolescent mothers. Adolescence 1998;33:565-74.
Houle JN. Depressive symptoms and all-cause mortality in a nationally representative longitudinal study with time-varying covariates. Psychosom Med 2013;75:297-304.
Lavretsky H, Alstein LL, Olmstead RE, Ercoli LM, Riparetti-Brown M, Cyr NS, et al.
Complementary use of tai chi chih augments escitalopram treatment of geriatric depression: A randomized controlled trial. Am J Geriatr Psychiatry 2011;19:839-50.
Cuijpers P, Vogelzangs N, Twisk J, Kleiboer A, Li J, Penninx BW, et al.
Differential mortality rates in major and subthreshold depression: Meta-analysis of studies that measured both. Br J Psychiatry 2013;202:22-7.
Breslow NE. Statistics in epidemiology: The case-control study. J Am Stat Assoc 1996;91:14-26.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3], [Table 4]