Achieving clinical response in postpartum depression leads to improvement in health-related quality of life

Abstract

Objective

To evaluate the health-related quality of life (HRQoL) burden associated with postpartum depression (PPD), determine the extent to which clinical response impacts HRQoL, and estimate the impact of PPD and clinical response on healthcare resource utilization (HRU) and productivity.

Methods

Patient data (n = 127) from two multicenter, randomized, double-blind, placebo-controlled phase 3 clinical trials evaluating the safety and efficacy of brexanolone injection in adults with PPD were employed for these posthoc analyses. HRQoL and health utility was assessed with the SF-36-v2 Health Survey (SF-36v2) acute version. The 17-item Hamilton Rating Scale for Depression (HAMD-17) total score was used to identify clinical response (≥50% reduction in HAMD-17 total score). Baseline HRQoL burden was assessed by comparison to age- and gender-adjusted population normative data from the 2009 QualityMetric PRO Norming study. The impact of clinical response was evaluated by comparing day 7 and day 30 SF-36v2 scores between clinical responders and non-responders. Interpretations of the meaningfulness of clinical response were indirectly estimated via 2017 National Health and Wellness Survey data linking SF-36v2 mental component summary (MCS) scores to (HRU) and productivity.

Results

Baseline HRQoL of patients with PPD was significantly below normative values. Day 7 and day 30 clinical response were associated with large and statistically significant improvements in HRQoL, greater likelihood of meeting SF-36v2 responder definitions, and reduced impairment. MCS levels corresponding to those observed in clinical responders were linked to lower HRU and productivity loss relative to non-responders.

Conclusions

PPD places a substantial burden on HRQoL. Achievement of rapid clinical response (at day 7) and clinical response sustained several weeks following the end of treatment (day 30) led to significant improvement in HRQoL, suggesting the importance of identifying women with PPD and providing effective treatment options.

Keywords:

• Postpartum depression
• health-related quality of life
• SF-36v2 Health Survey
• healthcare resource utilization
• productivity

Introduction

Postpartum depression (PPD) is one of the most common medical complications during and after pregnancy, affecting an estimated 13.2% of women giving birth in the United States and 17.7% globally^1–7. Up to 30% of women with PPD may have depressive symptoms persisting at least a year postpartum⁸. Suicide is strongly associated with depressive symptoms and is a leading cause of pregnancy-related mortality^9–16. Mothers with PPD may experience difficulties with functioning and bonding with their infants^17,18. The impact of PPD on offspring can be substantial, with long-term detrimental effects on cognitive, psychological and physical outcomes in children of women with PPD^19–22.

Fully capturing the burden of depression, its treatment, and actively evaluating the degree to which a new therapy alleviates this burden can be a challenge and must include more than only assessing changes in clinical symptoms. It has been shown that the impact of depressive clinical symptoms extends well beyond abnormalities in mood and neurovegetative symptoms to also include multiple aspects of psychological well-being, social and role function, and physical functioning^23–29. It is therefore crucial that clinical trials of new therapies for depression incorporate patient-reported outcomes (PROs) to systematically evaluate how treatment impacts patients’ overall functioning and well-being; PROs are increasingly used as endpoints in clinical trials^30–33.

One of the most widely used PROs for measuring patient functioning and well-being is the SF-36 Health Survey (SF-36). The SF-36 is a general health status instrument that measures eight domains of functional health status and well-being, including physical, social and role functioning, general health status, vitality and psychological well-being. The instrument was constructed for use in health policy evaluations, general population surveys, clinical research and practice, other applications involving diverse patient populations, and can be scored as a preference-based measure to provide utility data for economic evaluation^34,35. A systematic review of PROs used in randomized controlled clinical trials also found the SF-36 to be the most widely used instrument in registered clinical trials³⁰. Initial development and validation studies of the SF-36 confirmed the extensive health-related quality of life (HRQoL) burden associated with depression when scores were compared with those of other chronic diseases and the general population^36,37. Significant and meaningful improvements in SF-36 scale and summary measure scores have been shown among patients with major depressive disorder (MDD) with clinical response and remission, including those who were determined to be responders and remitters as defined by improvement in the 17-item Hamilton Depression Rating Scale (HAMD-17)^38–41. In these studies, responders and remitters showed significantly greater improvement in all or most SF-36 scales when compared to non-responders.

Similar to findings in MDD, PPD has been shown to produce significant impairment in many aspects of HRQoL as measured with the SF-36, including physical, social and role functioning, general health perceptions, vitality and psychological well-being^20,42–44. However, to our knowledge, no studies have investigated the impact of clinical response to PPD treatment on HRQoL. The present analysis sought to further define the HRQoL burden associated with PPD and to determine the extent to which a clinical response impacts HRQoL. Interpretation guidelines and indirect estimation of healthcare resource utilization and productivity loss were utilized to understand the meaningfulness of HRQoL burden and improvements associated with clinical response.

Methods

Sample

The data for this post-hoc analysis come from two multicenter, randomized, double-blind, placebo-controlled phase 3 clinical trials, conducted under breakthrough therapy designation, and designed to evaluate the safety and efficacy of brexanolone injection in adults with moderate-to-severe PPD (NCT02942004 and NCT02942017). Details of the trial design and patient samples are described elsewhere⁴⁵. Women were eligible to participate if they were aged 18–45 years, 6 months postpartum or less at the time of the study, with a diagnosis of moderate to severe PPD and a qualifying HAMD-17 total score (Study 1: ≥26; Study 2: 20‒25). Individuals were excluded from the study if they had active psychosis, medical history of schizophrenia, bipolar disorder, or schizoaffective disorder, had attempted suicide during the current episode of postpartum depression, or a history of alcohol or drug abuse in the previous 12 months. Participants were randomly assigned (1:1:1) to receive a single intravenous injection of either brexanolone 90 µg/kg/hr, brexanolone 60 µg/kg/hr, or placebo for 60 h in Study 1 or (1:1) brexanolone 90 µg/kg/hr or placebo for 60 h in Study 2. The trials were conducted under an umbrella protocol allowing for integrated analysis; the integrated efficacy dataset, which includes patients who were randomized to either the recommended brexanolone 90 µg/kg/hr (BRX) dose or placebo, was employed herein.

Outcome measures

The primary efficacy outcome measure in both trials was the least-squares mean difference in the HAMD-17 total score from baseline to 60 h. The HAMD-17 is a clinician-administered instrument to assess the severity of depression⁴⁶. Response (improvement in depression symptoms) was defined as having a 50% or greater reduction from baseline in HAMD-17 total score.

The SF-36v2 is a 36 item, self-report survey of functional health and well-being^35,47. In both brexanolone trials, the acute form of the SF-36v2 was administered, which has a 1-week recall period. Responses to 35 of the 36 items are used to compute an 8-domain profile of functional health and well-being scores. The 8-domain profile consists of the following subscales: Physical Functioning (PF), Role Limitations due to Physical Health (RP), Bodily Pain (BP), General Health Perceptions (GH), Vitality (VT), Social Functioning (SF), Role Limitations due to Emotional Problems (RE), and Mental Health (MH). Psychometrically-based physical and mental health component summary scores (PCS and MCS, respectively) are computed from the eight domain scores to give a broader metric of physical and mental HRQoL. The eight SF-36v2 scales and two summary measures are scored using norm-based methods that standardize the scores to a mean of 50 and a standard deviation of 10 in the general US population, with higher scores indicative of better health status. Minimal important difference thresholds to evaluate the meaningfulness of group mean differences, as well as patient-level responder definitions, have previously been defined for each SF-36v2 domain and the component scores⁴⁷. Lastly, a subset of items is used to derive 6 domains that are scored to produce a single health-state utility index, the SF-6D³⁴. Utility scores range from 0 (dead) to 1(full health); the SF-6D ranges from 0.301 to 1.

The SF-36v2 was added to the trials as an exploratory endpoint by protocol amendment, and therefore was completed by a subset of patients in the clinical trials. Both the HAMD-17 and SF-36v2 were administered at baseline, and days 7, 14, 21 and 30. The focus of these analyses will be on baseline and day 7 (the earliest post-treatment timepoint with both HAMD-17 and SF-36v2 assessment). Parallel analyses were also conducted on day 30, as the latest post-treatment timepoint available.

Statistical analyses

The analytic sample was comprised of all patients with available HAMD-17 and SF-36v2 at one or both of baseline or day 7. All analyses were conducted using all data available for the specific analysis and employing pooled data between the BRX and placebo study groups, in order to maximize the sample size available for analysis. Demographic and clinical characteristics for all clinical trial participants contributing any data were summarized using frequencies for categorical measures and means and standard deviations for continuous measures. Statistical significance was not adjusted for multiple comparisons.

Health-related quality of life burden of postpartum depression

General population normative data used to estimate the burden of PPD on HRQoL came from the internet-based 2009 QualityMetric PRO Norming Study⁴⁷. The sample was drawn from a national probability sample of US non-institutionalized adults drawn from the KnowledgePanelⁱ, which was maintained by Knowledge Networks. The data from the norming study were adjusted to the age and gender of the PPD trial sample using separate least squares multiple regression models for each SF-36v2 scale and summary measure score. Student’s t-tests were conducted to test for statistical significance of differences in SF-36v2 scores between the BRX trial sample and the adjusted general population norms. Cohen’s effect size for standardized differences was calculated to help interpret the magnitude of the difference in mean scores between samples⁴⁸. In addition, the percentage of patients with PPD who score meaningfully below the general population norm was determined. A patient’s baseline SF-36v2 score was considered below the norm if the score difference from the norm was greater than 1.96 standard errors of measurement (SEM), which represents the 95% confidence interval around an individual patient score. We hypothesized that patients with PPD would have SF-36v2 scale and summary measure scores significantly below the general population.

Health-related quality of life and clinical response

The impact of a clinical response on HRQoL was investigated in several ways to provide meaningful interpretations of outcomes. Clinical response was assessed at day 7 and day 30, and defined as a 50% improvement in HAMD-17 total score from baseline to the time of assessment. Day 7 was selected for evaluating the impact of a clinical response on HRQoL this was the earliest post-baseline timepoint with both HAMD-17 and SF-36v2 assessments available and therefore the data closest to the time point of the trials’ primary endpoint (60 h post-baseline). Day 30 analyses were also conducted to assess the longer-term impact of clinical response on HRQoL, as this was the latest post-treatment timepoint available (4 weeks post-treatment).

First, mean changes in SF-36v2 scale, summary measures and SF-6D scores from baseline to day 7 and from baseline to day 30 were compared within and between HAMD-17 responders and non-responders. Student’s t-tests were conducted to test for the statistical significance of the change from baseline to day 7 and day 30 and the differences in mean change scores between HAMD-17 responders and non-responders. Mean differences between these groups were interpreted in light of previously established minimal important difference thresholds. We hypothesized that HAMD-17 responders would show a significant change from baseline in SF-36v2 scale, summary measures and SF-6D scores at both day 7 and day 30, and significantly greater change as compared to non-responders. Analyses also were conducted on each of the SF-36v2 scales and summary measures and SF-6D using the responder definition established for each measure to examine group differences in the percentage of patients who showed a meaningful decline (change from baseline to day 7/day 30 ≤ −1*responder definition), meaningful improvement (change from baseline to day 7/day 30 ≥ responder definition), or no change (change from baseline to day 7/day 30 within ± responder definition)⁴⁷. Chi-square tests were conducted to assess whether the proportion of patients in each of these three SF-36v2 change score categories differed between HAMD-17 responder and non-responder groups. We hypothesized that a larger proportion of HAMD-17 responders would show meaningful improvement at both day 7 and day 30, as compared with non-responders.

The meaningfulness of clinical response was further evaluated by comparing mean SF-36v2 scores at day 7 and day 30 to age- and gender-adjusted population norms stratified by HAMD-17 response status, in order to assess the extent to which HAMD-17 responders would be more likely to achieve normal HRQoL levels compared to HAMD-17 non-responders. We hypothesized that mean SF-36v2 scores of HAMD-17 responders at both day 7 and day 30, would meet or exceed normal HRQoL levels, defined as being within the established minimal important difference thresholds of normative levels, whereas non-responders would not meet normal HRQoL levels.

To interpret the meaning of changes in HRQoL scores associated with clinical response in terms of their impact on patients’ daily functioning, responses to specific items of the SF-36v2 at baseline and day 7 as well as baseline and day 30 were compared between HAMD-17 responders and non-responders. Specifically, the percentages of patients whose item-level response indicated a high level of impairment (response options of “all of the time” or “most of the time”) at baseline and day 7/day 30 were compared for selected SF-36v2 scale items. All items selected for this analysis contribute to scales most related to mental health status (mental health, role emotional, social functioning and vitality), which were expected to be the most responsive to changes in depression status. We hypothesized that, despite similar levels of high impairment at baseline, a smaller percentage of HAMD-17 responders would show high impairment at both day 7 and day 30, as compared with non-responders.

Lastly, data from the 2017 National Health and Wellness Survey (NHWS) were used to indirectly estimate differences in the utilization of psychiatrist/psychologist services and work productivity based on SF-36v2 MCS scores at baseline, and as observed for responders and non-responders at day 7 and day 30⁴⁹. MCS was selected since this measure represents the four SF-36v2 scales most related to mental health status, where the largest response in scores was observed in this study. A subset of the NHWS data that matched the brexanolone trial sample in terms of age, gender, and depression diagnosis was used. We hypothesized that higher levels of MCS would correspond to less utilization of psychiatrist/psychologist services and missed work, and further that a greater proportion of HAMD-17 responders would reach higher MCS levels at day 7 and day 30 as compared to non-responders.

Results

Overall 127 patients were included in the analytic sample, with a mean age of 27 years and a mean baseline HAMD-17 score of 25 (Table 1). In total, 87 patients were white (67%), 34 (27%) were black or African-American, and 21 (17%) were Hispanic or Latino. Over 40% of patients had a history of depression (44%) or anxiety (43%). Prior history of PPD was present in 39% of patients (64% with a prior birth), and 28% had a family history of PPD. A higher proportion of patients had onset of PPD during four weeks postpartum (75%) than during the third trimester (25%) and 26% of patients were taking concomitant antidepressants at baseline.

Table 1. Baseline demographics and characteristics for clinical trial patients with PPD contributing to analytic sample (n = 127).

Characteristic	Mean (SD) or n (%)
Age in years, mean (SD)	26.96 (5.43)
Ethnicity
Hispanic or Latino, n (%)	21 (17%)
Not Hispanic or Latino, n (%)	106 (83%)
Race
Black or African-American, n (%)	34 (27%)
White, n (%)	86 (69%)
Other, n (%)	6 (5%)
Height in cm, mean (SD)	164.65 (7.14)
Weight in kg, mean (SD)	84.57 (22.47)
Body-mass index in kg/m², mean (SD)	31.14 (7.80)
Personal history of psychiatric disorders
Depression (non-PPD), n (%)	56 (44%)
Anxiety, n (%)	55 (43%)
Premenstrual dysphoric disorder, n (%)	2 (2%)
Substance abuse, n (%)	1 (1%)
Schizophrenia, n (%)	0 (0%)
Previous PPD episode, n (%)	49 (39%)
Previous PPD episode among parous women only (n = 77), n (%)	49 (64%)
Family history of PPD
Yes, n (%)	36 (28%)
No, n (%)	91 (72%)
Onset of PPD, n (%)
Third trimester, n (%)	32 (25%)
Within 4 weeks of delivery, n (%)	95 (75%)
Antidepressant use at baseline
Yes, n (%)	33 (26%)
No, n (%)	94 (74%)
Baseline depression symptoms
HAMD-17 total score*, mean (SD)	25.43 (3.56)
Randomized treatment
Brexanolone, n (%)	63 (50%)
Placebo, n (%)	64 (50%)

Notes. *Assessed before injection on day 1.

Demographic and clinical characteristics are summarized for all clinical trial participants contributing data to any analyses reported herein.

Data are n (%) or mean (SD).

Weight, height and body mass index data were measured at screening.

Abbreviations. HAMD-17, 17-item Hamilton Rating Scale for Depression; PPD, postpartum depression.

Health-related quality of life burden of postpartum depression

Comparisons with age- and gender-adjusted general population norms revealed that the HRQoL of patients with PPD was significantly below normative values for all SF-36v2 scales, the mental summary measure, and the SF-6D (Table 2). Differences were greatest in the SF-36v2 domains of vitality, social functioning, role emotional and mental health, and the mental health summary measure, with scores 1 to 1.5 standard deviations below normative values, all large effect size differences (Cohen’s d > 0.8). Furthermore, more than 80% of patients with PPD scored meaningfully below the age- and gender-adjusted norm on each of these scales. Mean score differences on the physical functioning, role physical, bodily pain and general health scales between patients with PPD and general population norms were in the small (0.2 > Cohen’s d < 0.5) to moderate effect size range (0.5 > Cohen’s d < 0.8) and more than one-third of patients with PPD scored meaningfully below the normative values across these physical health scales.

Table 2. Comparison of baseline SF-36v2 Health Survey Scale scores between patients with PPD and adjusted general population norms.

SF-36v2 scale	PPD sample			General population			Mean difference	p-Value^a	Cohen’s D Effect Size	% Meaningfully below norm^b
	n	Mean	SE	n	Mean	SE
Physical functioning	121	47.4	0.85	2059	52.3	0.410	−4.9	.004	−0.334	41.3
Role-physical	121	39.9	1.06	2057	52.4	0.420	−12.5	<.001	−0.791	74.4
Bodily pain	121	46.6	0.92	2056	51.7	0.430	−5.1	.004	−0.330	43.0
General health	121	45.8	0.75	2061	51.3	0.430	−5.5	.002	−0.362	37.2
Vitality	121	33.0	0.61	2057	48.3	0.440	−15.3	<.001	−1.027	84.3
Social functioning	121	28.8	0.81	2057	49.1	0.440	−20.3	<.001	−1.316	87.6
Role-emotional	121	23.9	0.94	2057	49.5	0.440	−25.6	<.001	−1.609	97.5
Mental health	121	28.4	0.72	2060	47.9	0.430	−19.5	<.001	−1.307	91.7
Physical summary	121	52.7	0.86	2056	53.2	0.420	−0.5	.761	−0.035	33.1
Mental summary	121	20.8	0.86	2056	47.1	0.450	−26.3	<.001	−1.660	96.7
SF-6D utility index	121	0.547	0.006	1997	0.750	0.010	−0.203	<.001	−0.635	87.6

Notes: ^aTwo-sided t-test for two independent sample means; ^bdefined as below the age- and gender-adjusted norm minus 1.96*(Standard Error of Measurement).

Abbreviations. PPD, postpartum depression; SE, standard error.

Health-related quality of life and clinical response

Patients with PPD with HAMD-17 response at day 7 showed large and statistically significant improvements in all SF-36v2 scales, the mental summary measure and the SF-6D as did those with the response at day 30 (Table 3). Excluding the physical summary measure, the mean improvement in scores ranged from 4.76 to 25.44 points on day 7 and from 6.20 to 29.56 on day 30, which are all larger than the minimal important change in scores established for these SF-36v2 scales and score improvements of this magnitude are in the moderate to the large effect size range. Scales most related to mental health status showed the greatest mean score improvements, approaching or exceeding 2 standard deviations on the vitality, social functioning, role emotional, mental health, and mental summary scales. Meaningful improvement in scores was also observed on many of the SF-36v2 scales and the SF-6D for day 7 and day 30 HAMD-17 non-responders. Specifically, the mean improvement on the role physical, vitality, social functioning, role emotional and mental health scales, the mental summary measure and the SF-6D all exceeded the minimal important change established for these scales at both day 7 and day 30, as well as physical functioning at day 30. Significant differences in mean score improvement were observed between HAMD-17 responders and non-responders on the SF-36 v2 bodily pain, general health, vitality, social functioning, role emotional and mental health scales, the mental summary measure and SF-6D at both day 7 and day 30. The mean differences between groups were most prominent on those SF-36v2 scales related to mental health status where mean differences were all in the large effect size range.

Table 3. Mean and categorical changes in SF-36v2 scale scores from baseline to day 7 and to day 30 by HAMD-17 clinical response status.

	n	Mean change	SD	Within group	Between group		Better^a		Same		Worse		Day 7		Day 30
				p-Value	Day 7 p-Value	Day 30 p-Value	n^b	%^c	n^b	%^c	n^b	%^c	χ²	p-Value	χ²	p-Value
Physical functioning					.069	.661							3.864	.145	0.436	.804
Day 7 responder	59	4.76	9.5	.0002			24	40.7	32	54.2	3	5.1
Day 7 non-responder	60	1.93	8.5	.0854			18	30.0	33	55.0	9	15.0
Day 30 responder	74	6.20	9.1	<.001			35	47.3	35	47.3	4	5.4
Day 30 non-responder	38	5.38	9.9	.0018			16	42.1	19	50.0	3	7.9
Role physical					.109	.518							0.111	.946	0.285	.867
Day 7 responder	59	8.99	11.4	<.001			36	61.0	17	28.8	6	10.2
Day 7 non-responder	60	5.62	11.4	<.001			35	58.3	18	30.0	7	11.7
Day 30 responder	74	11.21	12.3	<.001			49	66.2	20	27.0	5	6.8
Day 30 non-responder	38	9.68	11.2	<.001			27	71.1	9	23.7	2	5.3
Bodily pain					.007	<.001							4.481	.106	10.930	.004
Day 7 responder	59	6.14	8.2	<.001			28	47.5	28	47.5	3	5.1
Day 7 non-responder	60	1.50	10.0	.2509			22	36.7	28	46.7	10	16.7
Day 30 responder	74	8.33	9.3	<.001			47	63.5	23	31.1	4	5.4
Day 30 non-responder	38	0.73	10.0	.6565			12	31.6	20	52.6	6	15.8
General health					.004	.021							8.249	.016	7.690	.021
Day 7 responder	59	7.09	7.1	<.001			23	39.0	36	61.0	0	0.0
Day 7 non-responder	60	3.04	7.8	.0038			12	20.0	44	73.3	4	6.7
Day 30 responder	74	8.14	8.2	<.001			39	52.7	33	44.6	2	2.7
Day 30 non-responder	38	4.27	8.4	.0034			10	26.3	25	65.8	3	7.9
Vitality					<.001	<.001							24.905	<.001	19.194	<.001
Day 7 responder	59	18.91	11.4	<.001			52	88.1	6	10.2	1	1.7
Day 7 non-responder	60	6.12	8.3	<.001			27	45.0	30	50.0	3	5.0
Day 30 responder	74	19.93	11.7	<.001			64	86.5	10	13.5	0	0.0
Day 30 non-responder	38	7.45	11.5	<.001			19	50.0	16	42.1	3	7.9
Social functioning					<.001	<.001							16.929	<.001	12.076	.002
Day 7 responder	59	18.01	11.9	<.001			48	81.4	10	16.9	1	1.7
Day 7 non-responder	60	7.99	11.4	<.001			27	45.0	29	48.3	4	6.7
Day 30 responder	74	23.11	11.7	<.001			66	89.2	8	10.8	0	0.0
Day 30 non-responder	38	11.97	13.4	<.001			24	63.2	12	31.6	2	5.3
Role emotional					<.001	<.001							7.528	.023	3.443	.179
Day 7 responder	59	20.05	14.7	<.001			48	81.4	10	16.9	1	1.7
Day 7 non-responder	60	9.16	11.4	<.001			35	58.3	22	36.7	3	5.0
Day 30 responder	74	24.45	15.6	<.001			63	85.1	8	10.8	3	4.1
Day 30 non-responder	38	12.96	14.4	<.001			27	71.1	9	23.7	2	5.3
Mental health					<.001	<.001							12.165	.002	39.344	<.001
Day 7 responder	59	21.42	12.1	<.001			50	84.7	9	15.3	0	0.0
Day 7 non-responder	60	7.22	8.9	<.001			34	56.7	23	38.3	3	5.0
Day 30 responder	74	24.27	10.9	<.001			71	95.9	3	4.1	0	0.0
Day 30 non-responder	38	9.00	13.7	<.001			17	44.7	18	47.4	3	7.9
Physical summary					.985	.552							2.397	.302	0.766	.682
Day 7 responder	59	0.82	7.8	.4231			23	39.0	16	27.1	20	33.9
Day 7 non-responder	60	0.79	7.9	.4443			21	35.0	24	40.0	15	25.0
Day 30 responder	74	1.58	8.5	.1135			31	41.9	23	31.1	20	27.0
Day 30 non-responder	38	2.63	9.0	.0790			19	50.0	11	28.9	8	21.1
Mental summary					<.001	<.001							6.579	.037	15.800	<.001
Day 7 responder	59	25.44	15.3	<.001			53	89.8	6	10.2	0	0.0
Day 7 non-responder	60	9.74	9.8	<.001			43	71.7	16	26.7	1	1.7
Day 30 responder	74	29.56	14.1	<.001			69	93.2	5	6.8	0	0.0
Day 30 non-responder	38	12.40	15.7	<.001			25	65.8	9	23.7	4	10.5
SF-6D utility index					<.001	<.001							21.162	<.001	16.399	<.001
Day 7 responder	59	0.186	0.12	<.001			52	88.1	7	11.9	0	0.0
Day 7 non-responder	59	0.050	0.09	<.001			30	50.8	20	33.9	9	15.3
Day 30 responder	74	0.238	0.14	<.001			71	95.9	2	2.7	1	1.4
Day 30 non-responder	38	0.085	0.10	<.001			26	68.4	8	21.1	4	10.5

Notes. ^aDefined as significant change in SF-36 subscale using 80% confidence intervals; ^bn is defined where data are available for both HAMD-17 and SF-36 at both baseline and day 7, or baseline and day 30; ^c% defined as the row group (sum of row percentages is 100).

Abbreviation. HAMD-17, 17-item Hamilton Rating Scale for Depression.

In both groups, a greater percentage of patients showed better outcomes than worse outcomes across all SF-36v2 scales, summary measures, and the SF-6D (Table 3). Significant differences in the categories of change favoring the responders were observed on the general health, vitality, social functioning, role emotional and mental health scales, the mental summary measure, and SF-6D. For each of these scales, except the general health scale, more than 80% of responders had a better outcome, which was nearly double the percentage of better outcomes observed for the non-responders. On day 30, significant differences in the categories of change favoring the responders were observed on the bodily pain, general health, vitality, social functioning, and mental health scales, the mental summary measure, and SF-6D (Table 3).

Figure 1 presents mean SF-36v2 scores at day 7 and day 30 for both groups compared to age- and gender-adjusted norms. As shown, day 7 SF-36v2 scores were significantly higher among HAMD-17 responders compared to non-responders across all scales except the physical summary measure, and day 30 scores were higher across all scales. The differences in scores were more pronounced among scales most related to mental health status. Scores of clinical responders met (within the established minimal important difference thresholds) or exceeded adjusted normative values on the physical functioning, bodily pain, general health, vitality, and mental health scales and the physical summary measure at day 7 and met or exceeded normative values for all scales at day 30. Where significant differences were observed, the SF-36v2 scores of non-responders remained well below adjusted normative values, with the exception of physical functioning and the physical health component summary at day 30.

Figure 1. Comparison of day 7 (A) and day 30 (B) SF-36v2 Health Survey scale scores between patients with PPD stratified by clinical response status and age- and gender-adjusted general population norms. *p < .05. Abbreviations. PF, physical functioning; RP, role physical; BP, bodily pain; GH; general health; VT, vitality; SF, social functioning; RE, role emotional; MH, mental health; PCS, physical component summary; MCS, mental component summary; HAMD-17, 17-item Hamilton Rating Scale for Depression; PPD, postpartum depression

As a basis for interpreting the changes in HRQoL scores from baseline to day 7 and to day 30, the content of SF-36v2 items most related to mental health status was examined. As shown in Table 4, the percentage of HAMD-17 responders reporting “feeling tired” all or most of the time in the past week dropped from 88.3% at baseline to 29.0% at day 7. Comparatively, 93.4% of non-responders reported “feeling tired” all or most of the time in the past week at baseline, which only declined to 69.8% at day 7. Similar patterns were observed for being full of energy, health interfering with social activities, emotional health causing less productivity at work, feeling calm and peaceful, and feeling downhearted and depressed, with substantially larger declines in the proportion of patients showing high levels of impairment for responders as compared to non-responders. A similar pattern of results also was found at day 30 (Table 4).

Table 4. Item content-based interpretation of SF-36v2 scales most related to mental health status by HAMD-17 clinical response status.

% Who reported high impairment	Day 7 responder		Day 7 non-responder		Day 30 responder		Day 30 non-responder
	n^a	% High impairment^b	n^a	% High impairment^b	n^a	% High impairment^b	n^a	% High impairment^b
Feeling tired
Baseline	60	88.3	61	93.4	75	92.0	38	92.1
Follow-up (day7/30)	62	29.0	63	69.8	79	27.8	40	70.0
Full of energy^c
Baseline	60	88.3	61	85.2	75	86.7	38	89.5
Follow-up (day7/30)	62	24.2	63	68.3	79	25.3	40	60.0
Social activities interfered
Baseline	60	71.7	61	60.7	75	65.3	38	65.8
Follow-up (day7/30)	62	14.5	63	38.1	79	2.5	40	32.5
Cut down time spent on work
Baseline	60	60.0	61	75.4	75	64.0	38	76.3
Follow-up (day7/30)	62	6.5	63	34.9	79	5.1	40	35.0
Accomplish less at work
Baseline	60	76.7	61	77.0	75	78.7	38	73.7
Follow-up (day7/30)	62	14.5	63	47.6	79	6.3	40	45.0
Calm and peaceful^c
Baseline	60	81.7	61	78.7	75	80.0	38	81.6
Follow-up (day7/30)	62	17.7	63	60.3	79	10.1	40	65.0
Downhearted and depressed
Baseline	60	78.3	61	75.4	75	80.0	38	73.7
Follow-up (day7/30)	62	6.5	62	41.9	79	7.6	40	35.0

Notes: ^an defined as the column group denominator for % high impairment calculation; ^bDefined as the two most impaired levels of 5-level Likert scale; ^cImpairment means “feeling full of energy” or “feeling calm and peaceful” none or little of the time.

Abbreviation. HAMD-17, 17-item Hamilton Rating Scale for Depression.

Table 5 presents the link between day 7 and day 30 SF-36v2 MCS score categories and health care utilization and missed days at work as derived from the 2017 NHWS. At baseline, 73% of patients in the brexanolone trials had MCS scores corresponding to the lowest category (<25) with a mean score of 20.8. This MCS level in the NHWS sample of young women (ages 18–45) with depressive symptoms in the US corresponded to 42.3% having 1 or more visits to a psychiatrist/psychologist in a 6 month period, with a mean number of visits of 10.9, and over one-quarter (26.0%) missing one or more days at work in the past 7 days (mean hours missed = 5.62). On day 7, 56% of HAMD-17 responders had scores of 45 or higher, with a mean score of 46.3, corresponding in the NHWS sample to a category of MCS scores where only 24.1% of young women in the US reported 1 or more visits to a psychiatrist/psychologist in the last 6 months (mean number of visits among those with at least one = 7.77) and only 5.6% reported missing one or more days at work in the past 7 days (mean hours missed = 1.18). In contrast, only 10% of day 7 HAMD-17 non-responders had MCS scores of 45 or above. At day 30, 74.7% of HAMD-17 responders had MCS scores of 45 or higher, with a mean score of 50.0, corresponding in the NHWS sample to a category of MCS scores where only 21.0% of young women in the US reported at least one visit to a psychiatrist/psychologist in the last 6 months (mean number of visits among those with at least one = 5.91) and 7.4% missed one or more days of work in the past 7 days (mean hours missed = 1.45). Among HAMD-17 non-responders at day 30, 20.0% had MCS scores of 45 or above.

Table 5. Estimates of baseline, day 7, and day 30SF-36v2 mental health component summary score distributions and corresponding estimates of visits to psychiatrist or psychologist in the last 6 months and missed work days.

	Brexanolone Trial, MCS Distribution					National Health and Wellness Survey
SF-36v2 MCS Score Categories	Baseline, %	Day 7 HAMD-17 responder, %	Day 7 HAMD-17 non-responder, %	Day 30 HAMD-17 responder, %	Day 30 HAMD-17 non-responder, %	1 or more visits to psychiatrist/psychologist, %	Visits to psychiatrist/psychologist, mean (SD)	Hours missed at work in past 7 days	1 or more work days missed in past 7 days (%)
<25	72.7%	3.2%	31.7%	2.5%	35.0%	42.3%	10.91 (11.7)	5.62 (10.9)	26.0%
25 to <35	17.4%	17.7%	38.1%	3.8%	25.0%	31.5%	8.68 (9.0)	3.72 (7.3)	18.5%
35 to <45	9.1%	22.6%	20.6%	19.0%	20.0%	25.6%	7.25 (8.6)	2.30 (6.1)	11.4%
45 to <50	0.8%	12.6%	4.8%	19.0%	7.5%	24.1%	7.77 (12.8)	1.18 (3.9)	5.6%
≥50	0.0%	43.5%	4.8%	55.7%	12.5%	21.0%	5.91 (6.6)	1.45 (5.1)	7.4%

Abbreviations. MCS, mental health component summary; HAMD-17, 17-item Hamilton Rating Scale for Depression.

Discussion

The results of this post-hoc analysis suggest that patients with PPD had a significantly lower HRQoL level than the general population and that those patients who respond to treatment experienced rapid improvement in HRQoL to levels similar to those observed in the general population, indicative of reductions in healthcare resource utilization and productivity loss.

Overall, prior to treatment, patients with PPD reported substantially diminished functioning and well-being, compared to age- and gender-adjusted general population norms for the SF-36v2 Health Survey. The burden of PPD was most pronounced on SF-36v2 measures of mental health status, including psychological distress, role limitations due to emotional health, social functioning, and vitality. However, a substantial burden was also observed on measures related to physical health status. Comparisons with other conditions suggest that the HRQoL burden of PPD observed in this study should be considered clinically significant^36,37,50. For example, the profile of SF-36v2 scores was equal to or worse than that of conditions with well-known morbidities, such as MDD, heart disease, chronic migraine headaches, and rheumatoid arthritis^{24,36,37,50,51}. These findings help to fill a gap in evidence documenting the HRQoL burden associated with PPD²⁰.

Analyses conducted to evaluate the HRQoL impact of clinical response as derived from changes in HAMD-17 scores showed those patients who met the clinical response criteria also recovered rapidly as measured by HRQoL. On day 7, the mean improvement in SF-36v2 scores among responders was significant and the magnitude of change was in the moderate to the large effect size range. Analysis of categorical changes in SF-36v2 scores also showed a significantly greater proportion of HAMD-17 responders getting “better” by an amount determined to be clinically meaningful for nearly all SF-36v2 scales compared to non-responders. More importantly, the SF-36v2 scores of HAMD-17 responders approached or exceeded age- and gender-adjusted general population norms at day 7 while the scores of non-responders remained considerably below the norms. A similar pattern of results found at day 30 suggests improvements in HRQoL can be sustained beyond the end of treatment.

Since the interpretation of change in HRQoL scale scores is not always intuitive, another strategy used to interpret the impact of a clinical response on HRQoL was the analysis of item-level response data. For example, more than 75% of patients at baseline reported accomplishing less at work during the past 7 days on an item from the SF-36v2 role emotional scale. On day 7, the percentage of HAMD-17 responders who reported accomplishing less at work dropped to 14.5%, while 47.6% of HAMD-17 non-responders continued to report accomplishing less at work. Similar trends were observed on other SF-36v2 items evaluated.

Lastly, indirect estimates associated with SF-36 MCS score categories, healthcare utilization, and missed days at work (as derived from the 2017 NHWS) were used to interpret the meaningfulness of clinical response in this study. The magnitude of improvement in MCS scores observed in this study among HAMD-17 responders was indicative of significant reductions in the likelihood of visits to a psychologist/psychiatrist and missed time at work.

Clinical implications

The profound improvement in HRQoL observed for HAMD-17 responders relative to non-responders highlights the importance of effectively treating patients with PPD. Approximately half of women with PPD go undiagnosed and over 80% go untreated^52,53, suggesting the substantial pre-treatment HRQoL burden observed in this study is unaddressed in most women experiencing PPD. Given the evidence suggesting that PPD symptoms commonly persist beyond the first postnatal year and can substantially impact children and family^8,21, this lack of treatment represents a significant public health concern. Furthermore, data presented herein suggest that women with PPD who receive effective treatment can experience improvement in HRQoL, and that this can extend beyond the end of treatment (day 30, 4 weeks post-treatment). These improvements may translate into not only a substantial reduction in the humanistic burden of depression but also reductions in healthcare resource utilization and productivity loss.

Clinicians including primary care physicians, obstetricians and gynecologists, and pediatricians can significantly improve the identification of PPD by using standardized, validated screening tools, such as the Edinburgh Postnatal Depression Scale or Patient Health Questionnaire 9 during prenatal and postnatal clinical interactions following recommendations by the American College of Obstetricians and Gynecologists and the US Preventive Services Task Force^54,55. Once identified, management of the patient should follow a stepped-care approach, with treatment intensity matched to the severity of symptoms, the acuteness of presentation, and prior history of response to treatment^56,57.

A recent cost-effectiveness analysis indicates that treatment of adults with PPD symptoms with brexanolone is cost-effective, based on a United States health care payer perspective, over an 11-year time horizon, particularly for patients with severe symptoms⁵⁸. While this analysis accounted for improved health-related quality of life and cost savings from lower healthcare resource utilization associated with a reduction in PPD symptoms, it did not account for additional cost savings of reduced productivity loss, consideration of which would likely result in a lower cost-effectiveness ratio.

Limitations

A limitation of this study is that the data come from randomized controlled clinical trials. Patients enrolled in these studies may not be representative of the broader population of women with PPD. As such, the HRQoL burden observed in this study may overestimate the HRQoL burden in the broader PPD population. Due to the extent of HRQoL burden observed at baseline in this study, the generalizability of the impact of a clinical response on HRQoL to a broader PPD population may also be limited. Another limitation related to the indirect estimates of resource utilization and productivity by MCS score categories is the assumption of no effect modification in the post-partum population compared to the general population of child-bearing age women used in this study. Finally, the indirect estimates of the impact of clinical response on healthcare resource utilization and productivity would benefit from a prospective evaluation.

Conclusion

To date, few studies have systematically evaluated the HRQoL burden associated with PPD. The findings from this study demonstrated that PPD places a tremendous burden on patients’ HRQoL. Patients’ physical and mental health status was observed to be well below general population norms and equal to or worse than diseases with more well-known morbidity. These findings address an important evidence gap in understanding the humanistic burden of PPD. Further, it was demonstrated that clinical response had a significant impact on HRQoL outcomes. In 7 days, the HRQoL profile of patients who achieved a clinical response was restored to normative levels, and at day 30, four weeks after the end of treatment, patients with a clinical response remained at normative HRQoL levels. Together these findings suggest the importance of identifying women with PPD and providing them with effective treatment options.

Transparency

Declaration of funding

Sage Therapeutics, Inc. provided funding for this study.

Declaration of financial/other relationships

MEG, MYH, VB, SJK, and AE-L are employees of Sage Therapeutics, Inc. and own stock or stock options in the company. PH was an employee of Sage Therapeutics, Inc. at the time the study was conducted. SM-B reports personal fees from MedScape and grants from Sage Therapeutics, Inc., awarded to the University of North Carolina (Chapel Hill, NC, USA) during the conduct of the brexanolone injection clinical trials and grants from Janssen, PCORI, and the NIH outside the submitted work. MK, SA, and MF are employees of Optum, Acaster Lloyd Consulting, Ltd. and AMF Consulting, respectively, which were paid by Sage Therapeutics to conduct the research reported in this manuscript. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

MEG, MK, SA, MF, AEL conceptualized the study. MF conducted the data analysis. SMB and SJK collected the data. SMB, MYH, VB, PH, and SJK provided critical feedback on design and findings. MEG and MK developed the manuscript with input from all authors. All authors have reviewed and approved of the final manuscript.

Acknowledgements

None.

Notes

i now GfK Custom Research, LLC; http://www.knowledgenetworks.com/ganp/index.html