Brief psychosocial-behavioral intervention with antidepressant reduces post-stroke depression significantly more than usual care with antidepressant: Living Well with Stroke randomized controlled trial
Background and Purpose
Depression following stroke is prevalent, diminishing recovery and quality of life. Brief behavioral intervention, adjunctive to antidepressant therapy has not been well evaluated for long-term efficacy in those with post-stroke depression.
Methods
101 clinically depressed ischemic stroke patients within four months of index stroke were randomly assigned to an 8 week brief psychosocial-behavioral intervention plus antidepressant or usual care, including antidepressant. Primary endpoint was reduction in depressive symptom severity at 12 months following entry.
Results
Hamilton Rating Scale for Depression (HRSD) raw score in the intervention group was significantly lower immediately post-treatment (p = < 0.001) and at 12 months (p = 0.05) compared to controls. Remission (HRSD < 10), was significantly greater immediately post-treatment and at 12 months in the intervention group compared to usual care control. The mean percent decrease (47% ± 26% intervention versus 32% ± 36% control, p = .02) and the mean absolute decrease (-9.2 ± 5.7 intervention versus -6.2 ± 6.4 control, p = 0.023) in HRSD at 12 months were clinically important and statistically significant in the intervention group compared to control.
Conclusion
A brief psychosocial/behavioral intervention is highly effective in reducing depression in both the short and long term.
Methods
Design
Living Well with Stroke (LWWS) was a randomized treatment efficacy study with usual care control, comparing brief psychosocial/behavioral intervention plus antidepressant to usual care and antidepressant. 17 The study was approved by the University of Washington Human Subjects Division (IRB) for protection of human subjects.
The primary aim was to determine the effect of a nurse-delivered psychosocial-behavioral intervention on depression in community dwelling post-stroke patients. The primary hypothesis was that patients with PSD provided a 9 session problem-solving, pleasant events intervention in addition to antidepressant treatment would have significantly reduced depression severity (Hamilton Rating Scale for Depression (HRSD) at 12 months post-stroke compared to PSD patients with usual care, including antidepressant treatment.
The secondary aims were to describe the time course for reduction in depressive symptoms and to determine the intervention effect on secondary endpoints: limitations in ability (physical function), limitation in participation and overall stroke impact.
Participants
Participants were 101 patients within 4 months of an ischemic stroke, verified by CT or MRI, who screened positive for depressive symptoms, and whose diagnosis of clinical depression was verified by a diagnostic interview using DSM IV criteria.23 Participants were not excluded for prior or current treatment for depression.
Study Protocol
Depression Screening
Patients hospitalized in four acute care hospitals in Seattle, Washington were screened for depression with the 30 item Geriatric Depression Scale (GDS).22 The diagnosis of depression was validated by the Diagnostic Interview and Structured Hamilton (DISH) 23 in those who scored ≥11 on the GDS and consented to the full study.
Psychosocial/Behavioral Intervention and Usual Care
All participants were given written stroke recovery materials from the American Stroke Association, including information about depression. They completed a medication diary each of the first eight weeks of the trial. These diaries were either mailed to the research coordinator (usual care) with reminder telephone calls or brought to the counseling session (psychosocial-behavioral intervention). All participants saw their stroke care or primary care provider for ongoing medical care, including adjustment of antidepressant medication, as scheduled by that provider.
Those randomized to the psychosocial-behavioral intervention met with a study interventionist nine times over eight weeks. Sessions were focused on the individual; however a participant could opt to have a family member or informal caregiver join these sessions. Fifteen caregivers joined sessions and provided data. The brief psychosocial/behavioral intervention was adapted from the “Seattle Protocols”, shown to reduce disability associated with depression in Alzheimer's Disease.18
Although antidepressant treatment is often provided to PSD patients as a community standard, we reasoned that without behavioral change there would not be a long-lasting change from the initial mood elevation seen with antidepressants. We used language pertinent to stroke and taught participants to view depressive symptoms as observable and modifiable behaviors that are initiated and maintained by person-environment interactions. The treatment goal was to increase the level of pleasant social and physical activity in order to improve mood. Specific problem-solving approaches were taught and solutions to behavioral challenges were individualized to each person. The content of these sessions is more fully outlined in our design publication17 and in the treatment manual available from the first author.
Usual care arm
Participants in the usual care arm saw their stroke care or primary care provider as scheduled by that provider. When employed, antidepressant medication was prescribed and adjusted by the usual care provider. Twenty-five caregivers joined the stroke survivor in this arm. We did not design an arm with equivalent time and attention but without the depression content because the Seattle Protocol trials had already demonstrated that an ‘attention control’ as well as a waiting list control had significantly less improvement in depression than did the active intervention, similar to other problem-solving interventions in multiple populations.1719
Antidepressant recommendation
Selective serotonin reuptake inhibitor (SSRI) antidepressant treatment was an informal standard of care in this community at the inception of this study. Therefore, the participant's own provider prescribed the SSRI or an alternate antidepressant, based on his or her own assessment and recommendation provided through a letter sent to each participant's provider. The study psychiatrist (RV) recommended sertraline as an initial antidepressant choice due to its tolerability in the setting of medical illness and a relatively lower incidence of cardiovascular side effects.
Study timetable and assessments
A full baseline dataset including demographics, stroke characteristics, NIH stroke scale (NIHSS) score,20 Geriatric Depression Scale (GDS).21 Diagnostic interview (DISH), including the Hamilton Rating Scale for Depression (HRSD),22 Barthel Index,23 Stroke Impact Scale (SIS),24 and percent perceived recovery (Overall Stroke Impact) 24 was assessed at entry, but prior to randomization. Measures of depressive symptoms, stroke impact and perceived recovery were repeated at 9 weeks (immediately post intervention), 21 weeks post entry (roughly 6 months post-stroke) and 12 months and 24 months post-stroke. Functional and quality of life outcomes were conceptualized according to the World Health Organization (WHO) classification of persisting impairments, disabilities and handicaps from a variety of chronic illnesses.25 This multidimensional model of illness considers interacting elements of body, person, and social function, replacing the terms disability and handicap with more descriptive terms: limitations in activities and restrictions in participation. We measured limitation in physical ability by the Barthel Index, ADL, strength and mobility subscales of the SIS; limitation in participation by the communication and work/recreation subscales of the SIS, and overall stroke impact by percent perceived recovery from the SIS.
Randomization and masking
Randomization status was generated by a computerized adaptive randomization procedure after the method of Pocock and Simon.26 This ensured that the two groups remained balanced with respect to important predictors of outcomes: severity of stroke (NIHSS), severity of depression at baseline (HRSD), age and gender. All outcome assessors were masked to the participant's randomization status at each data collection point. We did not detect any breaches in masking.
Sample size and Statistical analysis
Over 1000 in-patients with stroke were tracked for potential screening eligibility. The majority were not eligible (comatose, not ischemic stroke, lived out of region, no evidence of sadness for example). Two hundred eighty nine agreed to be screened. One hundred forty eight of these were found eligible (GDS ≥11 and confirmed stroke), and 101 enrolled. (See Figure 1 for details on exclusions). The sample of 101 is adequate to detect a 0.5 standard deviation difference in average HRSD between the two groups. The primary endpoint was HRSD at 12 months. The intent to treat analysis controlled for baseline values of the primary outcome variable, using ANCOVA for continuous variables (HRSD) and logistic regression for binary variables (response or remission or not) at 12 months (primary end point), with alpha set at 0.05. Response was defined as ≥50% reduction in HRSD, consistent with prior literature. Remission has been defined variably as HRSD score of ≤9 (no longer meeting depression criterion)13, ≤7 (absence of any depressive symptoms)27 or ≤3 (equivalent to healthy controls).28 We used the more liberal HRSD score of ≤9 for comparison with previous studies of PSD.1314
Follow-up was achieved on all participants at 9 weeks, except the one person who dropped out of the study after one week. Two others in the intervention group dropped out between nine weeks and 12 months. Two participants in the intervention group died of underlying medical illness before 12 months; none died in the control group. Five dropped out or did not respond to 12 month follow-up in the control group (see Figure 1).
Design
Living Well with Stroke (LWWS) was a randomized treatment efficacy study with usual care control, comparing brief psychosocial/behavioral intervention plus antidepressant to usual care and antidepressant. 17 The study was approved by the University of Washington Human Subjects Division (IRB) for protection of human subjects.
The primary aim was to determine the effect of a nurse-delivered psychosocial-behavioral intervention on depression in community dwelling post-stroke patients. The primary hypothesis was that patients with PSD provided a 9 session problem-solving, pleasant events intervention in addition to antidepressant treatment would have significantly reduced depression severity (Hamilton Rating Scale for Depression (HRSD) at 12 months post-stroke compared to PSD patients with usual care, including antidepressant treatment.
The secondary aims were to describe the time course for reduction in depressive symptoms and to determine the intervention effect on secondary endpoints: limitations in ability (physical function), limitation in participation and overall stroke impact.
Participants
Participants were 101 patients within 4 months of an ischemic stroke, verified by CT or MRI, who screened positive for depressive symptoms, and whose diagnosis of clinical depression was verified by a diagnostic interview using DSM IV criteria.23 Participants were not excluded for prior or current treatment for depression.
Study Protocol
Depression Screening
Patients hospitalized in four acute care hospitals in Seattle, Washington were screened for depression with the 30 item Geriatric Depression Scale (GDS).22 The diagnosis of depression was validated by the Diagnostic Interview and Structured Hamilton (DISH) 23 in those who scored ≥11 on the GDS and consented to the full study.
Psychosocial/Behavioral Intervention and Usual Care
All participants were given written stroke recovery materials from the American Stroke Association, including information about depression. They completed a medication diary each of the first eight weeks of the trial. These diaries were either mailed to the research coordinator (usual care) with reminder telephone calls or brought to the counseling session (psychosocial-behavioral intervention). All participants saw their stroke care or primary care provider for ongoing medical care, including adjustment of antidepressant medication, as scheduled by that provider.
Those randomized to the psychosocial-behavioral intervention met with a study interventionist nine times over eight weeks. Sessions were focused on the individual; however a participant could opt to have a family member or informal caregiver join these sessions. Fifteen caregivers joined sessions and provided data. The brief psychosocial/behavioral intervention was adapted from the “Seattle Protocols”, shown to reduce disability associated with depression in Alzheimer's Disease.18
Although antidepressant treatment is often provided to PSD patients as a community standard, we reasoned that without behavioral change there would not be a long-lasting change from the initial mood elevation seen with antidepressants. We used language pertinent to stroke and taught participants to view depressive symptoms as observable and modifiable behaviors that are initiated and maintained by person-environment interactions. The treatment goal was to increase the level of pleasant social and physical activity in order to improve mood. Specific problem-solving approaches were taught and solutions to behavioral challenges were individualized to each person. The content of these sessions is more fully outlined in our design publication17 and in the treatment manual available from the first author.
Usual care arm
Participants in the usual care arm saw their stroke care or primary care provider as scheduled by that provider. When employed, antidepressant medication was prescribed and adjusted by the usual care provider. Twenty-five caregivers joined the stroke survivor in this arm. We did not design an arm with equivalent time and attention but without the depression content because the Seattle Protocol trials had already demonstrated that an ‘attention control’ as well as a waiting list control had significantly less improvement in depression than did the active intervention, similar to other problem-solving interventions in multiple populations.1719
Antidepressant recommendation
Selective serotonin reuptake inhibitor (SSRI) antidepressant treatment was an informal standard of care in this community at the inception of this study. Therefore, the participant's own provider prescribed the SSRI or an alternate antidepressant, based on his or her own assessment and recommendation provided through a letter sent to each participant's provider. The study psychiatrist (RV) recommended sertraline as an initial antidepressant choice due to its tolerability in the setting of medical illness and a relatively lower incidence of cardiovascular side effects.
Study timetable and assessments
A full baseline dataset including demographics, stroke characteristics, NIH stroke scale (NIHSS) score,20 Geriatric Depression Scale (GDS).21 Diagnostic interview (DISH), including the Hamilton Rating Scale for Depression (HRSD),22 Barthel Index,23 Stroke Impact Scale (SIS),24 and percent perceived recovery (Overall Stroke Impact) 24 was assessed at entry, but prior to randomization. Measures of depressive symptoms, stroke impact and perceived recovery were repeated at 9 weeks (immediately post intervention), 21 weeks post entry (roughly 6 months post-stroke) and 12 months and 24 months post-stroke. Functional and quality of life outcomes were conceptualized according to the World Health Organization (WHO) classification of persisting impairments, disabilities and handicaps from a variety of chronic illnesses.25 This multidimensional model of illness considers interacting elements of body, person, and social function, replacing the terms disability and handicap with more descriptive terms: limitations in activities and restrictions in participation. We measured limitation in physical ability by the Barthel Index, ADL, strength and mobility subscales of the SIS; limitation in participation by the communication and work/recreation subscales of the SIS, and overall stroke impact by percent perceived recovery from the SIS.
Randomization and masking
Randomization status was generated by a computerized adaptive randomization procedure after the method of Pocock and Simon.26 This ensured that the two groups remained balanced with respect to important predictors of outcomes: severity of stroke (NIHSS), severity of depression at baseline (HRSD), age and gender. All outcome assessors were masked to the participant's randomization status at each data collection point. We did not detect any breaches in masking.
Depression Screening
Patients hospitalized in four acute care hospitals in Seattle, Washington were screened for depression with the 30 item Geriatric Depression Scale (GDS).22 The diagnosis of depression was validated by the Diagnostic Interview and Structured Hamilton (DISH) 23 in those who scored ≥11 on the GDS and consented to the full study.
Psychosocial/Behavioral Intervention and Usual Care
All participants were given written stroke recovery materials from the American Stroke Association, including information about depression. They completed a medication diary each of the first eight weeks of the trial. These diaries were either mailed to the research coordinator (usual care) with reminder telephone calls or brought to the counseling session (psychosocial-behavioral intervention). All participants saw their stroke care or primary care provider for ongoing medical care, including adjustment of antidepressant medication, as scheduled by that provider.
Those randomized to the psychosocial-behavioral intervention met with a study interventionist nine times over eight weeks. Sessions were focused on the individual; however a participant could opt to have a family member or informal caregiver join these sessions. Fifteen caregivers joined sessions and provided data. The brief psychosocial/behavioral intervention was adapted from the “Seattle Protocols”, shown to reduce disability associated with depression in Alzheimer's Disease.18
Although antidepressant treatment is often provided to PSD patients as a community standard, we reasoned that without behavioral change there would not be a long-lasting change from the initial mood elevation seen with antidepressants. We used language pertinent to stroke and taught participants to view depressive symptoms as observable and modifiable behaviors that are initiated and maintained by person-environment interactions. The treatment goal was to increase the level of pleasant social and physical activity in order to improve mood. Specific problem-solving approaches were taught and solutions to behavioral challenges were individualized to each person. The content of these sessions is more fully outlined in our design publication17 and in the treatment manual available from the first author.
Usual care arm
Participants in the usual care arm saw their stroke care or primary care provider as scheduled by that provider. When employed, antidepressant medication was prescribed and adjusted by the usual care provider. Twenty-five caregivers joined the stroke survivor in this arm. We did not design an arm with equivalent time and attention but without the depression content because the Seattle Protocol trials had already demonstrated that an ‘attention control’ as well as a waiting list control had significantly less improvement in depression than did the active intervention, similar to other problem-solving interventions in multiple populations.1719
Antidepressant recommendation
Selective serotonin reuptake inhibitor (SSRI) antidepressant treatment was an informal standard of care in this community at the inception of this study. Therefore, the participant's own provider prescribed the SSRI or an alternate antidepressant, based on his or her own assessment and recommendation provided through a letter sent to each participant's provider. The study psychiatrist (RV) recommended sertraline as an initial antidepressant choice due to its tolerability in the setting of medical illness and a relatively lower incidence of cardiovascular side effects.
Study timetable and assessments
A full baseline dataset including demographics, stroke characteristics, NIH stroke scale (NIHSS) score,20 Geriatric Depression Scale (GDS).21 Diagnostic interview (DISH), including the Hamilton Rating Scale for Depression (HRSD),22 Barthel Index,23 Stroke Impact Scale (SIS),24 and percent perceived recovery (Overall Stroke Impact) 24 was assessed at entry, but prior to randomization. Measures of depressive symptoms, stroke impact and perceived recovery were repeated at 9 weeks (immediately post intervention), 21 weeks post entry (roughly 6 months post-stroke) and 12 months and 24 months post-stroke. Functional and quality of life outcomes were conceptualized according to the World Health Organization (WHO) classification of persisting impairments, disabilities and handicaps from a variety of chronic illnesses.25 This multidimensional model of illness considers interacting elements of body, person, and social function, replacing the terms disability and handicap with more descriptive terms: limitations in activities and restrictions in participation. We measured limitation in physical ability by the Barthel Index, ADL, strength and mobility subscales of the SIS; limitation in participation by the communication and work/recreation subscales of the SIS, and overall stroke impact by percent perceived recovery from the SIS.
Randomization and masking
Randomization status was generated by a computerized adaptive randomization procedure after the method of Pocock and Simon.26 This ensured that the two groups remained balanced with respect to important predictors of outcomes: severity of stroke (NIHSS), severity of depression at baseline (HRSD), age and gender. All outcome assessors were masked to the participant's randomization status at each data collection point. We did not detect any breaches in masking.
Sample size and Statistical analysis
Over 1000 in-patients with stroke were tracked for potential screening eligibility. The majority were not eligible (comatose, not ischemic stroke, lived out of region, no evidence of sadness for example). Two hundred eighty nine agreed to be screened. One hundred forty eight of these were found eligible (GDS ≥11 and confirmed stroke), and 101 enrolled. (See Figure 1 for details on exclusions). The sample of 101 is adequate to detect a 0.5 standard deviation difference in average HRSD between the two groups. The primary endpoint was HRSD at 12 months. The intent to treat analysis controlled for baseline values of the primary outcome variable, using ANCOVA for continuous variables (HRSD) and logistic regression for binary variables (response or remission or not) at 12 months (primary end point), with alpha set at 0.05. Response was defined as ≥50% reduction in HRSD, consistent with prior literature. Remission has been defined variably as HRSD score of ≤9 (no longer meeting depression criterion)13, ≤7 (absence of any depressive symptoms)27 or ≤3 (equivalent to healthy controls).28 We used the more liberal HRSD score of ≤9 for comparison with previous studies of PSD.1314
LWWS Flowchart
Follow-up was achieved on all participants at 9 weeks, except the one person who dropped out of the study after one week. Two others in the intervention group dropped out between nine weeks and 12 months. Two participants in the intervention group died of underlying medical illness before 12 months; none died in the control group. Five dropped out or did not respond to 12 month follow-up in the control group (see Figure 1).
Results
The two groups were comparable at baseline in both demographic and stroke variables, except for the prevalence of left hemisphere stroke and diabetes in the control group, as shown in Tables 1 and and2.2. The participants were predominantly male (59%), had moderate ischemic strokes, ranged in age from 25 to 88 years, and perceived themselves, on average, as about half recovered from their strokes (46.3% intervention; 55.3 % control). Roughly 70% of both groups had at least one episode of depression prior to this stroke and 60% of each group was taking antidepressants at entry. This increased to 77% in each randomization group during the 8 week active treatment period. Sertraline, citalopram, and paroxetine were the most commonly prescribed SSRI drugs. Tricyclic antidepressants were not commonly prescribed (11% of sample) with amitryptyline being the most common tricyclic antidepressant, prescribed at bedtime.
Table 1
Baseline Demographic Characteristics
| Characteristic | Intervention (N=48) | Control (N=53) |
|---|---|---|
| Gender | ||
| Male | 29 | 32 |
| Female | 19 | 21 |
| Age in years (mean, range) | 57 (25-88) | 57 (29-88) |
| Marital Status | ||
| Single | 7 | 8 |
| Married, Partnered | 19 | 22 |
| Widowed, divorced, separated | 22 | 23 |
| Current living arrangement | ||
| Alone | 4 | 3 |
| With spouse, partner, roomates | 22 | 24 |
| With other relatives | 13 | 17 |
| Group housing | 9 | 9 |
| Race, ethnicity | ||
| Hispanic | 3 | 2 |
| More than one race | 12 | 10 |
| White only | 29 | 35 |
| African-American only | 3 | 7 |
| Japanese only | 3 | 0 |
| Pacific Islander only | 0 | 1 |
| Korean only | 1 | 0 |
Table 2
Baseline Stroke and Health Characteristics
| Characteristic | Intervention (N=48) | Control (N=53) |
|---|---|---|
| NIHSSS (mean, SD, range) | 6.08 (4.4) 0-17 | 6.21 (5.05) 0-17 |
| HRSD (mean, SD, range) | 20.0 (4.53) 10-29 | 19.8 (4.15) 11-29 |
| Barthel Index (mean, SD) | 81.9 (23.2) | 83.5 (22.3) |
| Perceived percent recovery (mean, SD, range) | 46.3 (23.1) 0-90 | 55.3 (19.7) 10-100 |
| History of depression (%) | 75% | 69.8% |
| Currently taking antidepressant medication | 60.4% | 64.2% |
| Left hemisphere stroke | 37.5% | 52.8% |
| Right hemisphere stroke | 62.5% | 47.2% |
| Hypertensive medication | 81.3% | 75.5% |
| Heart Disease Congestive failure | 10.4% | 13.2% |
| Diabetes | 16.7% | 43.4% |
NIHSS = National Institutes of Health Stroke Scale
HRSD = Hamilton Rating Scale for Depression
Primary Endpoint
The primary hypothesis was supported. Mean decrease in HRSD in the intervention group was significantly greater at one year when compared with the control group (-9.2 ±5.7 intervention versus -6.2 ± 6.4 control, p = 0.023). As shown in Table 3a, when controlling for baseline HRDS scores, change in HRDS was significantly greater in the intervention group than in the control immediately post-treatment and at one year. The effect size is smaller after the first follow-up time but remains stable thereafter, favoring the intervention.
Table 3
| Table 3a. Mean change in Hamilton Rating Scale for Depression (HRSD) Over Time. Primary endpoint is shaded | ||||||||
|---|---|---|---|---|---|---|---|---|
| Follow-up Time | N,N | Intervention | Control | Difference | Confidence Interval | T | p-value | |
| 9 wk | 45,53 | -9.8 (4.9) | -3.6 (5.6) | -6.1 | (-8.2,-4.0) | -5.78 | <.001 | |
| 21 wk | 46,50 | -8.3 (6.8) | -6.0 (6.5) | -2.2 | (-4.8,0.4) | -1.67 | .098 | |
| 12 mo | 44,48 | -9.2 (5.7) | -6.2 (6.4) | -2.9 | (-5.4,-0.4) | -2.32 | .023 | |
| 24 mo | 34,33 | -11.3 (6.5) | -9.3 (4.7) | -2.2 | (-4.9,0.5) | -1.63 | .108 | |
| Table 3b. Percent of Participants in Remission (HRSD <= 9) Over Time. Primary end point is shaded. | ||||||||
| Follow-up Time | N,N | Intervention | Control | Odds Ratio | Confidence Interval | Chi-square | p-value | |
| 9 wk | 45,53 | 47% | 19% | 4.8 | (1,8,12.9) | 10.54 | .001 | |
| 21 wk | 46,50 | 46% | 22% | 3.4 | (1.3,8.7) | 6.98 | .008 | |
| 12 mo | 44,48 | 48% | 27% | 2.7 | (1.1,6.6) | 4.66 | .031 | |
| 24 mo | 34,33 | 65% | 46% | 2.3 | (0.8,6.7) | 2.29 | .130 | |
Further, as shown in Table 3b, significantly more of the intervention group was in remission (HRSD ≤9) immediately after treatment and at one year. The effect was strongest at the first follow-up and favored the intervention group at all time points. By 24 months, the control group remission rate approached the early response of the intervention group, while the intervention group continued to increase remission rate. There was no significant interaction effect on treatment response between treatment group and caregiver involvement at one year (data not shown). Over three quarters of each randomization group reported taking an antidepressant (SSRI, non-SSRI antidepressant or tricyclic) during the 8 week intervention period. There were no significant interactions of reported antidepressant use with either immediate or one year remission rates (data not shown).
Secondary Endpoints
There was no significant main effect for randomization group in examining indicators of limitation in ability (physical function), participation and percent recovery (overall impact) at 12 months. However, there was an interaction of remission status and scores for these variables, with significantly better scores for all indicators except the Barthel Index for those in remission (see Table 4). This finding is consistent with the suggestion that poorer rehabilitation and recovery is moderated by depressive symptoms.
Table 4
Activity, Participation and Stroke Impact by Remission Status at 12 months.
| Remission Status at 12 months | N | Mean (SD) | 95% CI for Mean | Significance | |
|---|---|---|---|---|---|
| Limitation in Ability | |||||
| 12m Strength (SIS) | No (HDRS >9) | 58 | 50.4(24.7) | 43.9, 56.9 | F = 14.65 p <0.001 |
| Yes (HDRS ≤9) | 34 | 71.1(25.7) | 62.2, 80.1 | ||
| 12m Mobility (SIS) | No(HDRS >9) | 58 | 62.5(24.1) | 56.2, 68.8 | F = 12.47 p = 0.001 |
| Yes (HDRS ≤9) | 34 | 80.8(23.8) | 72.5, 89.1 | ||
| 12m ADL (SIS) | No(HDRS >9) | 58 | 69.1(23.6) | 62.9, 75.3 | F = 8.57 p = 0.004 |
| Yes (HDRS ≤9) | 34 | 82.8(17.8) | 76.6, 89.0 | ||
| 12m Barthel Index | No(HDRS >9) | 58 | 86.4(19.7) | 81.2, 91.5 | F = 2.88 p = 0.09 |
| Yes (HDRS ≤9) | 34 | 93.1(15.7) | 87.6, 98.6 | ||
| Limitation in Participation | |||||
| 12m Communication | No(HDRS >9) | 58 | 79.3(19.4) | 74.2, 84.4 | F = 8.42 p = 0.005 |
| Yes (HDRS ≤9) | 34 | 90.7(5.5) | 85.2, 96.1 | ||
| 12m Work & Recreation | No(HDRS >9) | 58 | 53.0(24.5) | 46.5, 59.4 | F = 15.78 p <0.001 |
| Yes (HDRS ≤9) | 34 | 73.3(22.4) | 65.5, 81.2 | ||
| Overall Stroke Impact | |||||
| 12m % Recovered | No(HDRS >9) | 58 | 52.6(23.6) | 46.4, 58.8 | F =21.174 p <0.001 |
| Yes (HDRS ≤9) | 34 | 74.5(18.9) | 67.9, 81.1 | ||
CI = confidence interval, SIS = Stroke Impact Scale, ADL = Activities of daily living, All scales are 0-100 with higher scores being better
Primary Endpoint
The primary hypothesis was supported. Mean decrease in HRSD in the intervention group was significantly greater at one year when compared with the control group (-9.2 ±5.7 intervention versus -6.2 ± 6.4 control, p = 0.023). As shown in Table 3a, when controlling for baseline HRDS scores, change in HRDS was significantly greater in the intervention group than in the control immediately post-treatment and at one year. The effect size is smaller after the first follow-up time but remains stable thereafter, favoring the intervention.
Table 3
| Table 3a. Mean change in Hamilton Rating Scale for Depression (HRSD) Over Time. Primary endpoint is shaded | ||||||||
|---|---|---|---|---|---|---|---|---|
| Follow-up Time | N,N | Intervention | Control | Difference | Confidence Interval | T | p-value | |
| 9 wk | 45,53 | -9.8 (4.9) | -3.6 (5.6) | -6.1 | (-8.2,-4.0) | -5.78 | <.001 | |
| 21 wk | 46,50 | -8.3 (6.8) | -6.0 (6.5) | -2.2 | (-4.8,0.4) | -1.67 | .098 | |
| 12 mo | 44,48 | -9.2 (5.7) | -6.2 (6.4) | -2.9 | (-5.4,-0.4) | -2.32 | .023 | |
| 24 mo | 34,33 | -11.3 (6.5) | -9.3 (4.7) | -2.2 | (-4.9,0.5) | -1.63 | .108 | |
| Table 3b. Percent of Participants in Remission (HRSD <= 9) Over Time. Primary end point is shaded. | ||||||||
| Follow-up Time | N,N | Intervention | Control | Odds Ratio | Confidence Interval | Chi-square | p-value | |
| 9 wk | 45,53 | 47% | 19% | 4.8 | (1,8,12.9) | 10.54 | .001 | |
| 21 wk | 46,50 | 46% | 22% | 3.4 | (1.3,8.7) | 6.98 | .008 | |
| 12 mo | 44,48 | 48% | 27% | 2.7 | (1.1,6.6) | 4.66 | .031 | |
| 24 mo | 34,33 | 65% | 46% | 2.3 | (0.8,6.7) | 2.29 | .130 | |
Further, as shown in Table 3b, significantly more of the intervention group was in remission (HRSD ≤9) immediately after treatment and at one year. The effect was strongest at the first follow-up and favored the intervention group at all time points. By 24 months, the control group remission rate approached the early response of the intervention group, while the intervention group continued to increase remission rate. There was no significant interaction effect on treatment response between treatment group and caregiver involvement at one year (data not shown). Over three quarters of each randomization group reported taking an antidepressant (SSRI, non-SSRI antidepressant or tricyclic) during the 8 week intervention period. There were no significant interactions of reported antidepressant use with either immediate or one year remission rates (data not shown).
Secondary Endpoints
There was no significant main effect for randomization group in examining indicators of limitation in ability (physical function), participation and percent recovery (overall impact) at 12 months. However, there was an interaction of remission status and scores for these variables, with significantly better scores for all indicators except the Barthel Index for those in remission (see Table 4). This finding is consistent with the suggestion that poorer rehabilitation and recovery is moderated by depressive symptoms.
Table 4
Activity, Participation and Stroke Impact by Remission Status at 12 months.
| Remission Status at 12 months | N | Mean (SD) | 95% CI for Mean | Significance | |
|---|---|---|---|---|---|
| Limitation in Ability | |||||
| 12m Strength (SIS) | No (HDRS >9) | 58 | 50.4(24.7) | 43.9, 56.9 | F = 14.65 p <0.001 |
| Yes (HDRS ≤9) | 34 | 71.1(25.7) | 62.2, 80.1 | ||
| 12m Mobility (SIS) | No(HDRS >9) | 58 | 62.5(24.1) | 56.2, 68.8 | F = 12.47 p = 0.001 |
| Yes (HDRS ≤9) | 34 | 80.8(23.8) | 72.5, 89.1 | ||
| 12m ADL (SIS) | No(HDRS >9) | 58 | 69.1(23.6) | 62.9, 75.3 | F = 8.57 p = 0.004 |
| Yes (HDRS ≤9) | 34 | 82.8(17.8) | 76.6, 89.0 | ||
| 12m Barthel Index | No(HDRS >9) | 58 | 86.4(19.7) | 81.2, 91.5 | F = 2.88 p = 0.09 |
| Yes (HDRS ≤9) | 34 | 93.1(15.7) | 87.6, 98.6 | ||
| Limitation in Participation | |||||
| 12m Communication | No(HDRS >9) | 58 | 79.3(19.4) | 74.2, 84.4 | F = 8.42 p = 0.005 |
| Yes (HDRS ≤9) | 34 | 90.7(5.5) | 85.2, 96.1 | ||
| 12m Work & Recreation | No(HDRS >9) | 58 | 53.0(24.5) | 46.5, 59.4 | F = 15.78 p <0.001 |
| Yes (HDRS ≤9) | 34 | 73.3(22.4) | 65.5, 81.2 | ||
| Overall Stroke Impact | |||||
| 12m % Recovered | No(HDRS >9) | 58 | 52.6(23.6) | 46.4, 58.8 | F =21.174 p <0.001 |
| Yes (HDRS ≤9) | 34 | 74.5(18.9) | 67.9, 81.1 | ||
CI = confidence interval, SIS = Stroke Impact Scale, ADL = Activities of daily living, All scales are 0-100 with higher scores being better
Discussion
This trial is the first to report long term efficacy (one and two years) of a psychosocial treatment in combination with pharmacotherapy in PSD. It should be emphasized that the recruitment was not aimed at those with first-ever depression manifesting after stroke. Rather it was a treatment for stroke survivors who manifested a new or recurrent depression, consistent with the AIM trial described below.15 Few studies regarding the prevalence of PSD or its treatment describe prior history of depression, even though it is a predictor of future depression.1
This study extends the findings of two additional trials in stroke survivors that demonstrated short-term reduction in depression with structured care management 15 and with motivational interviewing. 16 The intervention in our trial is related to but not the same as those in these recently completed trials. LWWS was behavioral in focus, providing participants with education and skills training to combat their depression. The AIM trial had three components to care management: 1) an initial 20 minute structured psycho-education session with a nurse to activate the participant's interest and understanding of dealing with PSD and reduce stigma, 2) use of a medication algorithm to initiate a recommendation for an antidepressant to patient's primary provide and 3) a bimonthly telephone monitoring of depressive symptoms, medication side effects, and adherence. The timing and focus of sessions with the nurse counselor were substantially shorter than those in our study, and more focused on optimizing antidepressant therapy. While the participants in the treatment arm reduced depression scores significantly over the 12 week follow-up compared to the usual care arm, the drop in HRSD score and percent in remission were smaller than in our more behaviorally oriented study. This latter difference likely reflects the AIM trial's use of a lower remission score cut-off (HRSD score < 8). The AIM trial did not report longer term outcomes.15
Watkins and colleagues tested a motivational interviewing approach in a randomized controlled trial of 411 stroke inpatients, not limited to those with depressed mood. The goal was to assist patients still in hospital to recognize and achieving their own solutions to problems in adjustment to physical and emotional aspects of stroke. 16 Components of this intervention are similar to the problem-solving portions of our trial. Clinical psychologists provided four 30-60 minute individual sessions of motivational interviewing over 4 weeks to patients randomized to the treatment arm. A significantly larger proportion of those who received motivational interviews had ‘normal’ mood as measured by the General Hospital Questionnaire (GHQ-28) and did not endorse feeling sad or blue (depression screen) post- treatment compared to those with routine hospital care. Longer term protection from depressive symptoms was not reported.
We postulated that the antidepressant treatment would enable short-term mood elevation, while the behavioral treatment would provide cognitive restructuring and specific problem-solving skills for coping with ongoing stressors. Future planned analyses will explore the degree to which attainment of these skills is related to treatment response. While 77% of each group were prescribed and reported taking antidepressants during the 8 week treatment period, the doses and type of drug were not standardized. This is a limitation of the study but represents the context of everyday practice. Future planned subgroup analyses may help shed light on the role of antidepressants in the rate of remission and overall response to the behavioral treatment.
The findings of the LWWS study are similar to a British series that does not include PSD patients They have shown that a six session structured problem-solving psychological treatment can be delivered by various members of the primary healthcare team for a variety of emotional disorders, including major depression.293233 Recent comparison of antidepressant alone to combined problem-solving and antidepressant showed that both were equally effective in reducing depressive symptoms (HRSD) over time, with medication alone having a smaller reduction than the psychosocial or combined groups, as well as a sustained effect to one year.31 Community health nurses and general practitioners were equally effective, although patients were more satisfied with the community nurse treatment. 33
Psychosocial therapy combined with antidepressant has been significantly more successful in preventing recurrent depression in older adults than medication alone or psychosocial therapy and placebo. 34 More relevant to PSD are the trials of counseling and behavioral tailored problem-solving interventions for depressed medically ill elderly delivered by psychogeriatric teams and psychosocial nurse specialists in the home compared to “standard” care. These have shown a reduction in level of depression, improvement is self perceived health, and sometimes improved functional status. 3536 Our trial, in combination with the AIM trial and the motivational interviewing previously discussed add to our knowledge of the range of behavioral interventions effective in the PSD population. Further planned analyses of the LWWS trial will help suggest subgroups most responsive to treatment.
Summary
A brief psychosocial/behavioral intervention adjunctive to antidepressant therapy is highly effective in reducing depression and achieving remission in the short term, with the effect sustained for up to two years. Participants in the usual care control group also reduced their depression over the first year, but more slowly and with a lesser degree of remission. Those who achieved remission in either group had significantly better perceived recovery, lower limitation in ability and greater social participation at all time points.
Acknowledgments
This study was funded by the National Institute of Nursing Research, National Institutes of Health grant # R01NR007755 and registered with the clinical trials identifier: {"type":"clinical-trial","attrs":{"text":"NCT00194454","term_id":"NCT00194454"}}NCT00194454
http://www.clinicaltrials.gov/ct/show/{"type":"clinical-trial","attrs":{"text":"NCT00194454","term_id":"NCT00194454"}}NCT00194454?order=1
Abstract
Background and Purpose
Depression following stroke is prevalent, diminishing recovery and quality of life. Brief behavioral intervention, adjunctive to antidepressant therapy has not been well evaluated for long-term efficacy in those with post-stroke depression.
Methods
101 clinically depressed ischemic stroke patients within four months of index stroke were randomly assigned to an 8 week brief psychosocial-behavioral intervention plus antidepressant or usual care, including antidepressant. Primary endpoint was reduction in depressive symptom severity at 12 months following entry.
Results
Hamilton Rating Scale for Depression (HRSD) raw score in the intervention group was significantly lower immediately post-treatment (p = < 0.001) and at 12 months (p = 0.05) compared to controls. Remission (HRSD < 10), was significantly greater immediately post-treatment and at 12 months in the intervention group compared to usual care control. The mean percent decrease (47% ± 26% intervention versus 32% ± 36% control, p = .02) and the mean absolute decrease (-9.2 ± 5.7 intervention versus -6.2 ± 6.4 control, p = 0.023) in HRSD at 12 months were clinically important and statistically significant in the intervention group compared to control.
Conclusion
A brief psychosocial/behavioral intervention is highly effective in reducing depression in both the short and long term.
Depression is a serious sequelae of strokes of all types, affecting at least 33% of stroke survivors.1 Post-stroke depression (PSD) has been associated with poor recovery and rehabilitation response, 26 reduced social functioning and delayed return to work, 789, greater use of healthcare services, 10 and increased risk of subsequent cardiac and stroke events as well as increased mortality from all causes.21112
Antidepressants have demonstrated varying degrees of efficacy when tested over short follow-up periods. A recent Cochrane review concluded that while mood was improved in many antidepressant drug studies, there was no clear evidence of remission (no longer meeting entry depression criteria) or response (≥ 50% reduction in depression score) in the short or long term.13 In contrast a meta-analysis of 16 randomized controlled trials showed definite overall improvement in PSD in those taking a variety of antidepressants compared to placebo.14 Only recently have there been well-designed trials of non-pharmacologic treatments, both showing important reduction in PSD short-term (3months).1516 We believe ours is the first to report clinically and statistically important reduction in depression over the long term.
Footnotes
Conflict of Interest Disclosure: The above referenced grant was a source of funding for all authors. There are no other conflicts to disclose.
All authors participated in the design, conduct and analysis of the study. Dr. Mitchell wrote the initial draft of the paper, which was then reviewed and edited by the remaining authors. In addition, Ms. Buzaitis was lead therapist, Dr. Teri provided quality control supervision to the therapists, Dr. Veith provided psychiatric consultation to the therapists and Dr. Cain developed the randomization protocol and served as data and safety monitor.
Preliminary results were presented at the 2008 International Stroke Conference, with the abstract published in Stroke: Podium presentation ISC – published as: Mitchell, PH, Becker, KJ, Buzaitis, A, Cain, KC, Fruin, M, Kohen, R, Teri, L, Tirschwell, D & Veith, R. (2008) Brief psychosocial/behavioral intervention with antidepressant reduces post-stroke depression significantly more than antidepressant alone. Stroke: A Journal of the American Heart Association, 39 (2), 543.
Contributor Information
Pamela H. Mitchell, Department of Biobehavioral Nursing & Health Systems, University of Washington.
Richard C. Veith, Department of Psychiatry and Behavioral Science, University of Washington.
Kyra J. Becker, Department of Neurology, University of Washington.
Ann Buzaitis, Department of Biobehavioral Nursing and Health Systems, University of Washington.
Kevin C. Cain, Department of Biostatistics, University of Washington.
Michael Fruin, Department of Biobehavioral Nursing & Health Systems, University of Washington.
David Tirschwell, Department of Neurology, University of Washington.
Linda Teri, Department of Psychosocial and Community Health, University of Washington.
References
- 1. Hackett ML, Yapa C, Parag V, Anderson CSFrequency of depression after stroke: A systematic review of observational studies. Stroke. 2005;36:1330–1340.[PubMed][Google Scholar]
- 2. Morris PLP, Raphael B, Robinson RSClinical depression is associated with impaired recovery from stroke. Med J Austral. 1992;157:239–242.[PubMed][Google Scholar]
- 3. van de Weg FB, Kuik DJ, Lankhorst GJPost-stroke depression and functional outcome: A cohort study investigating the influence of depression on functional recovery from stroke. Clin Rehabil. 1999;13:268–272.[PubMed][Google Scholar]
- 4. Herrman N, Black SE, Lawrence J, Szekely C, Szalai JPThe Sunnybrook stroke study: A prospective study of depressive symptoms and functional outcome. Stroke. 1998;29:618–624.[PubMed][Google Scholar]
- 5. Ramasubbu R, Robinson RG, Flint AJ, Kosier T, Price TRFunctional impairment associated with acute poststroke depression: The stroke data bank study. J Neuropsychiatry Clin Neurosci. 1998;10:26–33.[PubMed][Google Scholar]
- 6. Ormel J, Kempen GI, Deeg DJ, Brilman EI, van Sonderen E, Relyveld JFunctioning, well-being, and health perception in late middle-aged and older people: Comparing the effects of depressive symptoms and chronic medical conditions. J Am Geriatr Soc. 1998;46:39–48.[PubMed][Google Scholar]
- 7. Parikh RM, Robinson RG, Lipsey JR, Starkstein SE, Fedoroff P, Price TRThe impact of poststroke depression on recovery in activities of daily living over a 2-year follow-up. Arch Neurol. 1990;47:785–789.[PubMed][Google Scholar]
- 8. Neau JP, Ingrand P, Mouille-Brachet C, Rosier MP, Couderq C, Alvarez A, Gil RFunctional recovery and social outcome after cerebral infarction in young adults. Cerebrovasc Dis. 1998;8:296–302.[PubMed][Google Scholar]
- 9. Goodwin RD, Devanand DPStroke, depression, and functional health outcomes among adults in the community. J Geriatr Psychiatry Neurol. 2008;21:41–46.[PubMed][Google Scholar]
- 10. Jia H, Damush TM, Qin H, Ried LD, Wang X, Young LJ, Williams LSThe impact of poststroke depression on healthcare use by veterans with acute stroke. Stroke. 2006;37:2796–2801.[PubMed][Google Scholar]
- 11. Everson SA, Roberts RE, Goldberg DE, Kaplan GADepressive symptoms and increased risk of stroke mortality over a 29-year period. Arch Intern Med. 1998;158:1133–1138.[PubMed][Google Scholar]
- 12. Morris PL, Robinson RG, Samuels JDepression, introversion and mortality following stroke. Aust N Z J Psychiatry. 1993;27:443–449.[PubMed][Google Scholar]
- 13. Hackett ML, Anderson CS, House AOManagement of depression after stroke: A systematic review of pharmacological therapies. Stroke. 2005;36:1098–1103.[PubMed][Google Scholar]
- 14. Chen Y, Guo JJ, Zhan S, Patel NCTreatment effects of antidepressants in patients with post-stroke depression: A meta-analysis. Ann Pharmacother. 2006;40:2115–2122.[PubMed][Google Scholar]
- 15. Williams LS, Kroenke K, Bakas T, Plue LD, Brizendine E, Tu W, Hendrie HCare management of poststroke depression: A randomized, controlled trial. Stroke. 2007;38:998–1003.[PubMed][Google Scholar]
- 16. Watkins CL, Auton MF, Deans CF, Dickinson HA, Jack CI, Lightbody CE, Sutton CJ, van den Broek MD, Leathley MJMotivational interviewing early after acute stroke: A randomized, controlled trial. Stroke. 2007;38:1004–1009.[PubMed][Google Scholar]
- 17. Mitchell PH, Teri L, Veith R, Buzaitis A, Tirschwell D, Becker K, Fruin M, Kohen R, Cain KCLiving well with stroke: Design and methods for a randomized controlled trial of a psychosocial behavioral intervention for poststroke depression. J Stroke Cerebrovasc Dis. 2008;17:109–115.[Google Scholar]
- 18. Teri L, Logsdon RG, McCurry SM. The Seattle protocols: Advances in behavioral treatment of Alzheimer's Disease. In: Grundman M, Feldman H, Fitten LJ, Winblad B, Giacobini E, editors. Research and practice in Alzheimer's Disease. Paris: Serdi Publisher; 2005. pp. 153–158. [PubMed]
- 19. Teri L, Logsdon RG, Uomoto J, McCurry SMBehavioral treatment of depression in dementia patients: A controlled clinical trial. J Gerontol B Psychol Sci Soc Sci. 1997;52:159–166.[PubMed][Google Scholar]
- 20. Lyden P, Brott T, Tilley B, Welch KM, Mascha EJ, Levine S, Haley EC, Grotta J, Marler J. Improved reliability of the NIH stroke scale using video training. NINDS tpa stroke study group. Stroke. 1994;25:2220–2226.[PubMed]
- 21. Yesavage JA, Brink TL, Rose TL, Lum O, Huang V, Adey M, Leirer VODevelopment and validation of a geriatric depression screening scale: A preliminary report. J Psychiatr Res. 1982;17:37–49.[PubMed][Google Scholar]
- 22. Freedland KE, Skala JA, Carney RM, Raczynski JM, Taylor CB, Mendes de Leon CF, Ironson G, Youngblood ME, Krishnan KR, Veith RCThe depression interview and structured hamilton (dish): Rationale, development, characteristics, and clinical validity. Psychosom Med. 2002;64:897–905.[PubMed][Google Scholar]
- 23. Collin C, Wade DT, Davies S, Horne VThe Barthel ADL index: A reliability study. Int Disabil Stud. 1988;10:61–63.[PubMed][Google Scholar]
- 24. Duncan PW, Bode RK, Min Lai S, Perera SRasch analysis of a new stroke-specific outcome scale: The stroke impact scale. Arch Phys Med Rehabil. 2003;84:950–963.[PubMed][Google Scholar]
- 25. Stucki G, Cieza A, Melvin JThe international classification of functioning, disability and health (ICF): A unifying model for the conceptual description of the rehabilitation strategy. J Rehabil Med. 2007;39:279–285.[PubMed][Google Scholar]
- 26. Pocock SJ, Simon RSequential treatment assignment with balancing for prognostic factors in the controlled clinical trial. Biometrics. 1975;31:103–115.[PubMed][Google Scholar]
- 27. Manber R, Kraemer HC, Arnow BA, Trivedi MH, Rush AJ, Thase ME, Rothbaum BO, Klein DN, Kocsis JH, Gelenberg AJ, Keller MEFaster remission of chronic depression with combined psychotherapy and medication than with each therapy alone. J Consult Clin Psychol. 2008;76:459–467.[Google Scholar]
- 28. Zimmerman M, Posternak MA, Chelminski IImplications of using different cut-offs on symptom severity scales to define remission from depression. Int Clin Psychopharmacol. 2004;19:215–220.[PubMed][Google Scholar]
- 29. Mynors-Wallis LProblem-solving treatment: Evidence for effectiveness and feasibility in primary care. Int J Psychiatry Med. 1996;26:249–262.[PubMed][Google Scholar]
- 30. Mynors-Wallis L, Davies I, Gray A, Barbour F, Gath DA randomised controlled trial and cost analysis of problem-solving treatment for emotional disorders given by community nurses in primary care. Br J Psychiatry. 1997;170:113–119.[PubMed][Google Scholar]
- 31. Mynors-Wallis LM, Gath DH, Day A, Baker FRandomised controlled trial of problem solving treatment, antidepressant medication, and combined treatment for major depression in primary care. BMJ. 2000;320:26–30.[Google Scholar]
- 32. Mynors-Wallis LM, Gath DH, Lloyd-Thomas AR, Tomlinson DRandomised controlled trial comparing problem solving treatment with amitriptyline and placebo for major depression in primary care. BMJ. 1995;310:441–445.[Google Scholar]
- 33. Kendrick T, Simons L, Mynors-Wallis L, Gray A, Lathlean J, Pickering R, Harris S, Rivero-Arias O, Gerard K, Thompson CCost-effectiveness of referral for generic care or problem-solving treatment from community mental health nurses, compared with usual general practitioner care for common mental disorders: Randomised controlled trial. Br J Psychiatry. 2006;189:50–59.[PubMed][Google Scholar]
- 34. Reynolds CF, 3rd, Frank E, Perel JM, Imber SD, Cornes C, Miller MD, Mazumdar S, Houck PR, Dew MA, Stack JA, Pollock BG, Kupfer DJNortriptyline and interpersonal psychotherapy as maintenance therapies for recurrent major depression: A randomized controlled trial in patients older than 59 years. JAMA. 1999;281:39–45.[PubMed][Google Scholar]
- 35. Banerjee S, Shamash K, Macdonald AJ, Mann AHRandomised controlled trial of effect of intervention by psychogeriatric team on depression in frail elderly people at home. BMJ. 1996;313:1058–1061.[Google Scholar]
- 36. Mossey JM, Knott KA, Higgins M, Talerico KEffectiveness of a psychosocial intervention, interpersonal counseling, for subdysthymic depression in medically ill elderly. J Gerontol A Biol Sci Med Sci. 1996;51:M172–178.[PubMed][Google Scholar]