A total of 1191 participants were recruited (583 started in the control condition and 608 started in the intervention condition) (Fig. 1) into the trial. All participants allocated to the intervention condition received the counseling intervention. Outcome data were available for 471 control participants (19% lost to follow-up) and 543 intervention participants (11% lost to follow-up) at 3-month follow-up. Reasons for lost to follow-up (at 3-month) included the participant moving outside the trial district (n = 25,14.0%), long-term stay out of district for work or other reasons (n = 90, 50.80%), major illness (n = 7, 4.0%), refusal to participate in the follow-up assessment (n = 47, 26.6%) and other reasons (n = 8, 4.5%). For the intervention condition, 8 months of follow-up data for period 5 was not collected due to logistical constraints (Table S1). At the 8-month follow-up, 416 participants started in the control condition and 276 participants started in the intervention condition. At follow-up periods, no participants displayed suicidal ideation, suicidal attempts, or self-harm.
Participant demographic characteristics at baseline are shown in Table 1. Compared to intervention participants, participants in the control condition were older, less educated, and more likely to be employed (Table 1).
Table 1
Baseline demographic characteristics of the participants by condition
Demographic variables
|
Control condition (n = 583)
|
Intervention condition (n = 608)
|
Age (years), Median (IQR)
|
45 (18)
|
40 (18)
|
Sex, N (%)
Male
|
249 (42 ∙8)
|
241 (39 ∙6)
|
Education (years of schooling)
Median (IQR)
|
3 (8)
|
5 (9)
|
Marital Status, N (%)
Currently married and living together
Married but separated
Divorced
Widowed
Never married
|
451 (77 ∙3)
10 (1 ∙7)
8 (1 ∙4)
83 (14 ∙1)
31 (5 ∙5)
|
407 (66 ∙8)
31 (5 ∙1)
8 (1 ∙3)
112 (18 ∙4)
51 (8 ∙4)
|
Employment status, N (%)
Unemployed
|
146 (25)
|
206 (33 ∙8)
|
Religion, N (%)
Hindu
Muslim
|
551(94 ∙5)
24 (4 ∙1)
|
595 (98)
11 (1 ∙9)
|
Caste, N (%)
Upper Caste
Other Backward Caste
Scheduled Tribes
Scheduled Caste
Don’t know
|
135 (23 ∙1)
239 (40 ∙9)
8 (1 ∙4)
144 (24 ∙7)
50 (8 ∙6)
|
168 (27∙7)
314 (51∙8)
6 (1)
111 (18∙3)
6 (1)
|
Data are presented as total number (%); IQR Interquartile range |
Baseline covariates were adjusted in the final regression analysis as participants’ mean scores for all scales at baseline (GHQ, PHQ, SRQ, GAD, QOL, WHODAS) were significantly higher in the intervention condition compared to control (Table 2). Average GHQ scores were on par or above the recommended cut-off of 6 for primary-care attendees in India24, and these community participants were not seeking any treatment in primary care. The mean score on the PHQ was 12 (SD = 5.6) and 14.2 (SD = 5.2) for control and intervention conditions respectively, indicating moderate depression. Mean scores on the GAD were 10 (SD = 5) and 11.6 (SD = 4.3) for the control and intervention conditions respectively, indicating moderate levels of anxiety. The average SRQ scores were high among both the intervention (mean = 14, SD = 4.5) and control condition participants (mean = 15, SD = 4) indicating a possible clinical diagnosis of a mental disorder.
Table 2
Baseline differences between control and intervention condition participants
Scales
|
|
Control condition
(n = 583)
|
Intervention condition
(n = 608)
|
P value
|
General health questionnaire
(GHQ-12)
|
Mean (SD)
|
6 ∙0 ± 2 ∙6
|
7 ∙7 ± 2 ∙7
|
0 ∙000
|
Median (Q1, Q3)
|
6 (4, 8)
|
8 (5, 10)
|
Patient Health Questionnaire
(PHQ-9)
|
Mean (SD)
|
12 ∙0 ± 5 ∙6
|
14 ∙2 ± 5 ∙2
|
0 ∙000
|
Median (Q1, Q3)
|
12 (8, 16)
|
14 (10, 18)
|
Self-reporting questionnaire
(SRQ-20)
|
Mean (SD)
|
14 ∙0 ± 4 ∙5
|
15 ∙0 ± 4 ∙0
|
0 ∙00
|
Median (Q1, Q3)
|
15 (11, 18)
|
16 (12, 180
|
Generalized anxiety disorder
(GAD-7)
|
Mean (SD)
|
10 ∙0 ± 5 ∙0
|
11 ∙6 ± 4 ∙3
|
0 ∙00
|
Median (Q1, Q3)
|
10 (7, 13)
|
12 (8, 15)
|
Social Participation scale (SPS)
|
Mean (SD)
|
20 ∙1 ± 18 ∙8
|
21 ∙5 ± 17 ∙8
|
0 ∙026NS
|
Median (Q1, Q3)
|
15 (6, 29)
|
18 (7, 31)
|
Quality of life (EQ-5D)
|
Mean (SD)
|
0 ∙64 ± 0 ∙15
|
0 ∙58 ± 0 ∙16
|
0 ∙000
|
Median (Q1, Q3)
|
0 ∙66 (0 ∙54, 0 ∙75)
|
0 ∙62 (0 ∙50, 0 ∙69)
|
WHO-Disability questionnaire (WHO-DAS)
|
Mean (SD)
|
24 ∙2 ± 8 ∙2
|
25 ∙7 ± 9 ∙04
|
0 ∙00
|
Median (Q1, Q3)
|
24 (18, 30)
|
25 (19, 31)
|
W1
(no ∙ of days)
|
Mean (SD)
|
18 ∙2 ± 10 ∙5
|
22 ∙3 ± 8 ∙9
|
0 ∙000
|
Median (Q1, Q3)
|
20 (10, 30)
|
30 (15, 30)
|
W2
(no ∙ of days)
|
Mean (SD)
|
4 ∙8 ± 7 ∙0
|
6 ∙5 ± 9 ∙0
|
0 ∙000
|
Median (Q1, Q3)
|
2 (0, 7)
|
3 (0, 10)
|
W3
(no ∙ of days)
|
Mean (SD)
|
5 ∙7 ± 6 ∙9
|
8 ∙4 ± 9 ∙2
|
0 ∙000
|
Median (Q1, Q3)
|
4 (0, 9)
|
5 (0, 12)
|
Values are presented as mean ± SD; median (Q1, Q3); non-parametric Mann-Whitney Test was used to test significance ∙ NS: Not significant |
W1: Overall, in the past 30 days, how many days were these difficulties (WHODAS scale) present? W2: In the past 30 days, for many days were you totally unable to carry out your usual activities or work because of any health conditions? W3: In the past 30 days, not counting the days that you were totally unable, for how many days did you cut back or reduce your usual activities or work because of any health condition? |
Table 3 provides regression estimates for both primary and secondary outcomes using Generalized Linear Mixed Models. After adjusting for baseline covariates, time points (periods), and cluster effects, categorical GHQ-12 recovery (a score of less than 3 at follow-up) was significantly higher among participants in the intervention condition at the end of 3 months (adjusted odds ratio OR 2.2; 95% CI 1.2 to 4.6). These effects were sustained at 8 months (adjusted OR 3.0; 95% CI 1.6 to 5.9). Similarly, continuous GHQ-12 scores were significantly lower among participants in the intervention condition at the end of 3 months (adjusted mean difference (AMD): -0.1; 95% CI: -1.6 to -0.3; effect size 0.2) and 8 months (AMD: -2.3; 95% CI: -3.1 to -1.5; effect size 0.5) (Table 3).
Table 3
Effect of intervention for primary and secondary outcomes
|
Control condition (n = 583)
|
Intervention condition (n = 608)
|
Regression estimates (95% CI)
|
P-value
|
ICC (95%CL)
|
Primary outcome
|
Number
|
Number
|
Adjusted odds ratio
|
|
|
CMD participants improved at 3 months
|
76
|
102
|
2∙2 (1∙2 to 4∙6)
|
0∙01
|
0∙1 (0∙06 to 0∙3)
|
CMD participants improved at 8 months
|
77
|
125
|
3∙0 (1∙6 to 5∙9)
|
0∙001
|
0.2 (0.07 to 0.3)
|
GHQ-12 on continuous scale
|
Median (Q1, Q3)
|
Median
(Q1, Q3)
|
Adjusted mean difference
(95% CI)
|
P value
|
Effect
size
|
GHQ-12 at 3 months
|
6 (3, 8)
|
5 (3, 8)
|
-0 ∙1 (-1 ∙6 to -0 ∙3)
|
0 ∙004
|
0 ∙2
|
GHQ-12 at 8 months
|
6 (3, 9)
|
3 (0, 6)
|
-2 ∙3 (-3 ∙1 to -1 ∙5)
|
0 ∙000
|
0 ∙5
|
Secondary outcome
at 3 months
|
Median (Q1, Q3)
|
Median
(Q1, Q3)
|
Adjusted mean difference
(95% CI)
|
P value
|
Effect size
|
Patient Health questionnaire (PHQ-9)
|
11 (7, 15)
|
11 (6, 15)
|
-2 ∙2 (-3 ∙2 to -1 ∙2)
|
0 ∙001
|
0 ∙3
|
Self-reporting questionnaire (SRQ-20)
|
14 (10, 18)
|
14 (8, 17)
|
-1 ∙5 (-2 ∙4 to -0 ∙5)
|
0 ∙001
|
0 ∙2
|
Generalized anxiety disorder (GAD-7)
|
9 (6, 13)
|
8 (5, 12)
|
-1 ∙5 (-2 ∙4 to -0 ∙5)
|
0 ∙001
|
0 ∙2
|
Social Participation scale (SPS)
|
14 (5, 26)
|
13 (3, 27)
|
-0 ∙8 (-3 ∙6 to 1 ∙9)
|
0.55 NS
|
0.04
|
Quality of life (EQ-5D)
|
0 ∙64(0 ∙50, 0 ∙74)
|
0 ∙68 (0 ∙57, 0 ∙76)
|
0 ∙05 (0 ∙02 to 0 ∙08)
|
0 ∙00
|
0 ∙3
|
WHO-Disability questionnaire (WHO-DAS)
|
23 (18, 29)
|
22 (16, 28)
|
-1 ∙3 (-2 ∙6 to 0 ∙01)
|
0.05
|
0.14
|
W1 (no ∙ of days)
|
15 (10, 30)
|
15 (10, 30)
|
-0 ∙8 (-2 ∙6 to 1 ∙0)
|
0.39 NS
|
0.06
|
W2 (no ∙ of days)
|
2 (0, 6)
|
1 (0, 7)
|
0 ∙07 (-0 ∙9 to 1 ∙1)
|
0.89 NS
|
0.05
|
W3 (no ∙ of days)
|
4 (1, 10)
|
5 (0, 10)
|
-0 ∙4 (-1 ∙8 to 0 ∙8)
|
0.48 NS
|
0.05
|
Secondary outcome at 8 months
|
Median (Q1, Q3)
|
Median (Q1, Q3)
|
Adjusted mean difference
(95% CI)
|
P value
|
Effect size
|
Patient Health questionnaire (PHQ-9)
|
12 (7 ∙7, 16)
|
6 (2, 11)
|
-4 ∙5 (-5 ∙8 to -3 ∙2)
|
0 ∙001
|
0 ∙6
|
Self-reporting questionnaire (SRQ-20)
|
14 (10, 18)
|
0 (0, 10)
|
-3 ∙4 (-4 ∙6 to -2 ∙0)
|
0 ∙001
|
0 ∙4
|
Generalized anxiety disorder (GAD-7)
|
10 (5, 14)
|
0 (0, 5)
|
-3 ∙6 (-4 ∙7 to -2 ∙4)
|
0 ∙001
|
0 ∙5
|
Social Participation scale (SPS)
|
14 (3, 31)
|
7 (0, 22)
|
-5 ∙2 (-8 ∙4 to -1 ∙9)
|
0 ∙002
|
0 ∙3
|
Quality of life (EQ-5D)
|
0 ∙67 (0 ∙56, 0 ∙75)
|
0 ∙73 (0 ∙59, 0 ∙84)
|
0 ∙06 (0 ∙03 to 0 ∙1)
|
0 ∙00
|
0 ∙3
|
WHO-Disability questionnaire (WHO-DAS)
|
23 (18, 29)
|
17 (13, 26)
|
-3 ∙5 (-5 ∙0 to -2 ∙0)
|
0 ∙001
|
0 ∙4
|
W1 (no ∙ of days)
|
15 (10 ∙ 30)
|
10 (3, 30)
|
-3 ∙2 (-5 ∙9 to -0 ∙4)
|
0 ∙02
|
0 ∙2
|
W2 (no ∙ of days)
|
3 (0, 10)
|
0 (0, 5)
|
-3 ∙0 (-4 ∙5 to -1 ∙4)
|
0 ∙000
|
0 ∙3
|
W3 (no ∙ of days)
|
5 (0, 10)
|
4 (0, 10)
|
-3 ∙1 (-4 ∙9 to -1 ∙3)
|
0 ∙001
|
0 ∙3
|
Adjusted estimates are presented as Odds ratio for primary outcome on binary scale and adjusted mean difference for secondary outcomes on continuous scales using a generalised linear mixed model and are adjusted for baseline covariates, random effect of clustering and fixed effect of underlying temporal trends. Analysis is also adjusted for demographic variables like age and gender, marital status and religion. NS: Not significant |
W1: Overall, in the past 30 days, how many days were these difficulties (WHODAS scale) present? W2: In the past 30 days, for many days were you totally unable to carry out your usual activities or work because of any health conditions? W3: In the past 30 days, not counting the days that you were totally unable, for how many days did you cut back or reduce your usual activities or work because of any health condition? |
We performed several additional model extensions. First, when allowing for different time points across clusters, the estimated treatment effect increased (Supplementary Table S2) and the correlation between observations in different periods was slightly lower than between observations within the same time points (period), indicating a slight decay of the correlation structure. Second, when testing for treatment effect heterogeneity, the intervention estimate increased compared to the base model, with similar correlations between observations in the intervention condition and control condition. (Supplementary Table S2). There were no significant differences (p < 0.05) when allowing for interaction between intervention and time period (Supplementary Table S3).
Secondary outcomes:
Identification and treatment of depression and anxiety symptoms
The champions’ case detection rate (accurate identification of common mental disorders) as compared to actual GHQ-12 scores (cut-off score of 3 or more) was 58% for all the clusters and periods in the intervention condition.
Data on counseling sessions attended was available for 89% (N = 543) of the intervention sample. Participants attended a mean of 5.2 (SD = 1.2) and a median of 6 (Q1, Q3: 5, 6) sessions. At the 3-month follow-up, 2% (n = 11) of participants were referred to public health services (either the PHC or the DMHP) by the champions. In the control group, 52% (n = 244) were similarly referred to public health services.
PHQ-9 depression scores were significantly lower among intervention condition participants compared to those in the control condition, with an adjusted mean difference (AMD) of -2.2 (95%CI: -3.2 to -1.2; effect size 0.3) at 3 months and AMD of -4.5 (95% CI: -5.8 to -3.2; effect size 0.6) at 8 months (p < 0.05). PHQ-9 scores among intervention condition participants changed from moderate depression scores of 10–12 at baseline to mild depression scores of 5–9 (range of adjusted means) at 8-month follow-up (p < 0.05). On anxiety symptoms as measured by the GAD-7,the intervention group showed significant improvement (AMD: -1.5 (-2.4 to -0.5; effect size 0.2) at 3 months and − 3.6 (-4.7 to -2.4; effect size 0.5) at 8 months follow-up (p < 0.05). Among participants in the intervention condition, GAD scores changed from moderate anxiety symptoms at baseline to mild at 3-month follow-up and diminished entirely at 8-month follow-up. There was no significant improvement at the 3-month follow-up on functioning scores, as assessed by the WHO-DAS (AMD of -3.5 (95%CI: -5 to -2; effect size 0.4). P WHO-DAS scores increased significantly at 8 months follow-up in intervention participants with AMD of -3.2 (-5.9 to -0.4 effect size 0.2), -3 (-4.5 to -1.4; effect size 0.3) and − 3.1 (-4.9 to -1.3; effect size 0.3) for W1, W2, W3 respectively. Compared to baseline, SRQ scores improved at 3 month (AMD − 1.5 (95%CI: -2.4 to -0.5; effect size 0.2) and at 8-month follow-up (AMD: -3.4 (95%CI: -4.6 to -2.0; effect size 0.4).Health-related quality of life assessed by EQ-5D, showed significant improvement with an AMD of 0.05 (95%CI: 0.02 to 0.08; effect size 0.3) at 3 months and 0.06 (95%CI: 0.03 to 0.1; effect size 0.3) at 8 month follow-up (Table 3).
On the Client Satisfaction Questionnaire, the average score was 24.0 ± 5.2 demonstrating a high satisfaction level with Atmiyata intervention at the end of 8 months.
Considering the loss to follow-up rate was greater than 10% at the end of 3 months follow-up and 8 months follow-up, baseline measures between the participants who continued the intervention and those lost to follow-up in each condition were analyzed separately. For data at 3 months follow-up in the intervention condition, no significant differences were observed between the baseline values on the GHQ-12, PHQ-9, GAD-7, and SRQ-20) among participants who remained in the intervention compared to those lost to follow-up. In the control condition, the baseline values were significantly higher among participants who remained in the intervention (p < 0.05).
At 8-month follow-up, mean baseline values of measures were similar (p > 0.05) in those who remained in the trial versus those lost to follow-up in both control and intervention conditions.
Moreover, the power to detect observed differences between the intervention and control condition remained at 80% at the end of 3 months. For an 8-month follow-up, data were considered until period 4 so based on the observed difference with the sample size of 276 in intervention and 217 in control with 3 steps and 48 clusters the power remained 80%.
The overall results across primary and secondary outcomes demonstrate that the intervention was effective, with effects sustained and improving over time, including at 8 months follow-up after counseling sessions ended.