Examining patient feedback and the role of cognitive arousal in treatment non-response to digital cognitive-behavioral therapy for insomnia during pregnancy.

Abstract

Objective.

Insomnia affects over half of pregnant and postpartum women. Early evidence indicates that cognitive-behavioral therapy for insomnia (CBTI) improves maternal sleep and mood. However, standard CBTI may be less efficacious in perinatal women than the broader insomnia population. This study sought to identify patient characteristics in a perinatal sample associated with poor response to CBTI, and characterize patient feedback to identify areas of insomnia therapy to tailor for the perinatal experience.

Participants.

Secondary analysis of 46 pregnant women with insomnia symptoms who were treated with digital CBTI in a randomized controlled trial.

Methods.

We assessed insomnia, cognitive arousal, and depression before and after prenatal treatment, then 6 weeks postpartum. Patients provided feedback on digital CBTI.

Results.

Residual cognitive arousal after treatment was the most robust factor associated with treatment non-response. Critically, CBTI responders and non-responders differed on no other sociodemographic or pretreatment metrics. After childbirth, short sleep (<6 hrs/night) was associated with maternal reports of poor infant sleep quality. Patient feedback indicated that most patients preferred online treatment to in-person treatment. Although women described digital CBTI as convenient and helpful, many patients indicated that insomnia therapy would be improved if it addressed sleep challenges unique to pregnancy and postpartum. Patients requested education on maternal and infant sleep, flexibility in behavioral sleep strategies, and guidance to manage infant sleep.

Conclusions.

Modifying insomnia therapy to better alleviate refractory cognitive arousal and address the changing needs of women as they progress through pregnancy and early parenting may increase efficacy for perinatal insomnia.

Name:

Insomnia and Rumination in Late Pregnancy and the Risk for Postpartum Depression

URL:

clinicaltrials.gov

Registration:

NCT03596879

Introduction

Over half of pregnant women meet diagnostic criteria for insomnia disorder (D⊘rheim et al., 2012) or endorse clinically significant insomnia symptoms (Kalmbach, Cheng, Sangha, et al., 2019; Mindell et al., 2015). Despite well-documented perinatal complications associated with insomnia, such as depression, suicidal ideation, and reduced quality of life (D⊘rheim et al., 2012; Manber et al., 2013; Mourady et al., 2017; Palagini et al., 2019; Volkovich et al., 2016), empirically supported treatment options are currently limited for pregnant women. No prescription or over-the-counter sleep medications have demonstrated safety and efficacy in large randomized controlled trials (RCTs) (Okun et al., 2015). However, emerging evidence from recent RCTs support cognitive and behavioral intervention as a viable treatment option for women with insomnia during pregnancy (Manber et al., 2019).

In the broader insomnia patient population, cognitive-behavioral therapy for insomnia (CBTI) is first-line treatment (Morgenthaler et al., 2006; Qaseem et al., 2016). Recent RCTs in perinatal samples support CBTI efficacy in the perinatal population. When delivered in-person, CBTI enhanced with psychoeducation on maternal and infant sleep produces superior treatment effects on perinatal insomnia as compared with control (Manber et al., 2019). However, few US adults with insomnia have access to behavioral sleep medicine specialists who provide CBTI treatment (Koffel et al., 2018; Thomas et al., 2016), thereby creating the need for efficacious remote treatment. We are not aware of any published trials testing CBTI via telemedicine in pregnant women. However, two digital CBTI RCTs in perinatal women have been published to date.

Felder and colleagues compared digital CBTI to wait-list control (Felder et al., 2020). Results showed that 44.0% of digital CBTI patients remitted based on Insomnia Severity Index ≤ 7 criterion, whereas just 22.3% of controls remitted. In a separate RCT, our team compared digital CBTI and digital sleep education control in women during late pregnancy (Kalmbach, Cheng, O’Brien, et al., 2020). Consistent with Felder et al.’s findings, we showed that digital CBTI, relative to sleep education control, alleviated insomnia symptoms and increased self-reported nightly sleep duration during pregnancy. After childbirth, we found that CBTI patients continued to sleep longer and were less likely to endorse sleep maintenance symptoms as compared with controls, suggesting that women with CBTI experience may be at least partially protected against postpartum sleep issues.

Although these RCT results are promising, we also observed a number of shortcomings of CBTI for perinatal insomnia. Rates of treatment response and insomnia remission are higher in non-pregnant insomnia patients who receive digital CBTI as compared with pregnant insomnia patients who receive digital CBTI. Using the same digital CBTI program and identical response and remission criteria, our lab previously showed that digital CBTI produced a response rate of 71% and remission rate of 54% in a non-pregnant insomnia sample (Cheng et al., 2018). By comparison, the two extant digital CBTI RCTs in pregnant women produced a treatment response rate of 37% and remission rates of 35-44% (Felder et al., 2020; Kalmbach, Cheng, O’Brien, et al., 2020). Higher rates of treatment response and remission among non-pregnant insomnia patients, relative to pregnant patients, likely indicate that CBTI must be tailored for the prenatal experience.

Moreover, women in our RCT reported that sleep symptoms changed significantly after childbirth (Kalmbach, Cheng, O’Brien, et al., 2020). As maternal sleep deprivation, poor infant sleep, and nighttime infant waking and feedings are common in early postpartum, it is unsurprising that short sleep became more common and sleep maintenance difficulties persisted, while sleep latency issues decreased in the newborn period. This change in sleep profiles reflects an evolution in clinical needs from pregnancy to early postpartum. Even though prenatal treatment protected against early postpartum sleep loss and sleep maintenance problems, 54% of postpartum women in the CBTI group reported short sleep of ≤ 6 hrs at night six weeks after childbirth (vs 91% of control patients who endorsed postnatal short sleep). In other words, postpartum women with insomnia report high levels of short sleep in early postpartum, even after showing improvement with prenatal insomnia therapy.

Although CBTI is highly efficacious across patient populations and health comorbidities (Cheng et al., 2018; Taylor & Pruiksma, 2014; Wu et al., 2015), standard CBTI may be sub-optimal for perinatal patients. We know of no RCTs that have tested in-person, unmodified CBTI in pregnant women. Rather, the only published in-person CBTI RCT for prenatal insomnia included psychoeducation on maternal and infant sleep for both active and control conditions. Although this modified CBTI produced remission rate of 64% (Manber et al., 2019)—which is consistent with RCT data from in-person CBTI trials in non-pregnant patient populations (Drake et al., 2018; Harvey et al., 2014; Morin et al., 2009)—over half of control patients also remitted during pregnancy. Notably, a >50% insomnia remission rate for controls is higher than typically reported in CBTI RCTs. While the control condition targeted arousal with pseudo-desensitization, which may have alleviated sleep symptoms for some patients, this hypothesis could not be tested directly.

Psychoeducation on maternal and infant sleep—during a period of great sleep upheaval—may have had some therapeutic benefit in both the CBTI and control conditions (Manber et al., 2019). Indeed, it is possible that psychoeducation on maternal and infant sleep may have contributed in part to higher remission rates for in-person CBTI (64%) compared with digital CBTI (35-44%) (Felder et al., 2020; Kalmbach, Cheng, O’Brien, et al., 2020). Tailoring insomnia therapy to meet the unique and changing needs of women as they progress through pregnancy and early parenting will likely improve the efficacy of CBTI for expecting and new moms.

While tailoring insomnia therapy to the perinatal period may enhance treatment response in this population, identifying other potential barriers to treatment response is also necessary to optimize insomnia therapy for pregnant and postpartum women going forward. In previous in-person and digital CBTI RCTs in the broader insomnia patient population, refractory cognitive arousal has been identified as a barrier to treatment response (Cheng et al., 2020; Kalmbach, Cheng, Arnedt, et al., 2019; Ong et al., 2009). Importantly, alleviating cognitive arousal symptoms helps facilitate CBTI effects on insomnia and depression outcomes (Cheng et al., 2020). Troublingly, pregnant women report pathogenic levels of cognitive arousal, particularly among those with insomnia symptoms (Kalmbach, Cheng, Ong, et al., 2020; Swanson et al., 2011). Unfortunately, in a prior RCT, CBTI did not alleviate perinatal cognitive arousal symptoms relative to control (Kalmbach, Cheng, O’Brien, et al., 2020). Even so, it remains unclear whether refractory cognitive arousal symptoms are associated with poor insomnia therapy response in this population.

This present study is a secondary analysis of 46 pregnant women with insomnia symptoms who were treated with digital CBTI in a RCT (Kalmbach, Cheng, O’Brien, et al., 2020). The first goal of this study was to identify patient characteristics associated with poor response to CBTI and with short sleep duration in early postpartum. We operationalized treatment response as a reduction of ≥ 6 points or more on the Insomnia Severity Index (ISI). Consistent with prior CBTI RCTs non-perinatal insomnia patients, we hypothesized that refractory cognitive arousal would be associated with treatment resistance (Cheng et al., 2020; Kalmbach, Cheng, Arnedt, et al., 2019; Ong et al., 2009). Specifically, we predicted that, after treatment, patients who did not respond to CBTI treatment would report higher levels of cognitive arousal (in the forms of nocturnal cognitive arousal, depressive rumination, and perseverative thinking) as compared with CBTI responders. In our previous trial, we showed that CBTI during pregnancy may protect against short sleep after childbirth. Notably, however, in-lab data have linked nocturnal cognitive arousal to short sleep duration (Kalmbach, Buysse, et al., 2020). Therefore, we also examined whether short sleep in postpartum was associated with cognitive arousal symptoms. In addition, we explored other patient factors that may be associated with non-response and short sleep, including pretreatment clinical symptoms (e.g., depression, snoring) and sociodemographic characteristics (e.g., age, race, poverty status).

The second study goal was to describe patient feedback on digital CBTI to better understand patient impressions on the benefits and limitations of standard CBTI for insomnia therapy in pregnancy and early postpartum. Patients provided quantitative and qualitative feedback regarding the positive and negative experiences of digital CBTI in pregnancy and early postpartum.

Materials and Methods

Study design

This study was approved by the Institutional Review Board at the Henry Ford Health System in Detroit, Michigan, USA (IRB #12204). All patients provided written consent to participate. The trial was registered at the US National Institutes of Health (ClinicalTrials.gov #NCT03596879). In a 6-hospital healthcare system, we enrolled patients nearing or entering the 3^rd trimester of pregnancy in a randomized controlled trial (RCT) comparing the efficacy of digital CBTI versus digital sleep education control. As this report is secondary analysis of factors associated with CBTI response and feedback on this digital intervention, we analyzed data from women randomized to CBTI.

Patients.

Invitations advertising a study on perinatal sleep (without mentioning either that we were focused on poor sleep or that we were evaluating sleep treatments) were sent via email and phone calls to 3585 pregnant patients in the health system. Inclusion criteria were clinically significant insomnia symptoms on a validated self-report instrument (Insomnia Severity Index score ≥ 10 (Morin et al., 2011), see Measures below)) and gestational age between 25 and 30 weeks at time of eligibility screening. Exclusion criteria included high risk pregnancy per self-report (e.g., pre-eclampsia, vasa previa, age >40), being in the care of the maternal-fetal medicine team for high risk pregnancy per electronic medical records, multiple pregnancy, prescription or over-the-counter sleep aid use at the time of screening, alcohol or recreational drug use at time of screening, shift work (nights, rotating), epilepsy or seizures, bipolar disorder, diagnosis of a sleep disorder that is untreated (other than insomnia), and severe depression (Edinburgh Postnatal Depression Scale score ≥ 19 (Cox et al., 1987), see Measures below).

A total of 535 women responded to our advertisements with interest in our study. Of these patients, 272 women consented to participating in the study, 267 of whom provided sufficient data for full eligibility determination, which were collected between September 12, 2018 and March 9, 2019. Of these 267 screeners, 156 women met insomnia symptom inclusion criteria for the RCT. However, 65 of these women were not enrolled into the trial for the following reasons (note: some women met multiple exclusion criteria): completed eligibility screening after gestational age limit for RCT inclusion (n=39), high risk pregnancy (most cases were considered high risk due to age; n=10), sleep medications (only doxylamine and diphenhydramine reported; n=10), severe depression (n=3), did not want to participate (n=2), shift work (n=2), patient-reported untreated sleep apnea diagnosis (n=2), multiple pregnancy (n=2), and current alcohol or recreational substance use (n=2). See Figure 1 for study flowchart.

Figure 1.

Flowchart of study enrollment and participation.

Study interventions

Enrolled patients were randomly assigned to receive either digital CBTI or digital sleep education control. For randomization, we used blocked randomization with a block size of 10 and an allocation ratio of 1:1. Research assistants uninvolved in hypothesis generation or data analysis generated randomization allocation, enrolled patients, and assigned patients to interventions. Due to the nature of the interventions, patients were not blinded to intervention, but they were not informed as to whether their treatment was considered active or control. For full details on the control condition, please refer to supplementary material.

Digital cognitive behavioral therapy for insomnia.

Patients randomized to CBTI completed the Sleepio program via the internet (www.sleepio.com, Big Health Inc.). In the present study, patients were granted access to Sleepio until either (A) they completed six sessions of digital CBTI or (B) they gave birth. The intervention covered behavioral sleep strategies (sleep restriction, stimulus control), cognitive components (e.g., cognitive restructuring, paradoxical intention), progressive muscle relaxation, and sleep hygiene. Only one modification was made for our patients: in sleep restriction, time in bed could not be prescribed as < 6 hrs. Sessions were directed by an animated virtual therapist (“the Prof”) who guides progress with the patient based on submitted sleep data.

Study outcomes

All study outcomes were assessed via online surveys hosted by Qualtrics. Data collection schedule included assessments at (1) pretreatment: 1-2 weeks before treatment, (2) posttreatment: a week after completing treatment or after discontinuing treatment during pregnancy, and finally (3) postnatal follow-up: 6 weeks after childbirth. Pretreatment measures administered during the screening process retained their original assessment windows, whereas pretreatment measures administered immediately before beginning treatment were modified to assess symptoms over ‘the prior week.’ Similarly, posttreatment and follow-up measures were modified to assess symptoms over ‘the prior week’ to minimize likelihood of administering posttreatment assessments after childbirth (note: all posttreatment assessments in the present study were completed before childbirth). Further, the uniform assessment period minimized assessment overlap with treatment (e.g., the Pittsburgh Sleep Quality Index measures sleep disturbance over the prior month (Buysse et al., 1989), thus posttreatment assessment would overlap with > 50% of the time patients were engaged in the 6-week treatment regimen).

Insomnia Severity Index (ISI) measured global insomnia symptom severity (Bastien et al., 2001; Morin et al., 2011). Pretreatment ISI was first administered as part of the screening process. Scores range from 0 to 28 with higher scores indicating greater severity. ISI ≥ 10 indicates clinically significant insomnia symptoms (Morin et al., 2011). Per standard practice, treatment response was operationalized as a reduction of ≥ 6 points on the scale between pre and posttreatment.

Global sleep disturbance was measured using the Pittsburgh Sleep Quality Index (PSQI) (Buysse et al., 1989) and was first assessed during screening. The PSQI measures a wide range of sleep parameters over the previous month including sleep duration, sleep latency, sleep aid use, and sleep difficulties related to insomnia, breathing difficulties, environmental stimuli, and other factors. A global cutoff score of PSQI > 5 differentiates poor sleepers from good sleepers. In addition to measuring global sleep disturbance, we used the PSQI to assess habitual sleep duration (item #4a). Short sleep was operationalized as ≤ 6 hrs.

Edinburgh Postnatal Depression Scale (EPDS) measured depression (Cox et al., 1987), which was first administered during screening. Scores range from 0 to 30 with higher scores indicating greater severity. EPDS scores ≥ 10 suggest clinically significant minor or major depression (Matthey et al., 2006). EPDS scores ≥ 19 represent severe depressive symptoms (McCabe-Beane et al., 2016), which was an exclusion criteria in the study.

Presleep Arousal Scale - Cognitive factor (PSAS-C) (Nicassio et al., 1985) measures tendency for experiencing cognitive arousal while trying to fall asleep at night. Pretreatment nocturnal cognitive arousal was assessed during screening. Example items from the PSAS-C are “review or ponder events of the day” and “can’t shut off thoughts.” Scores range from 8 to 40 with higher scores indicating greater nocturnal cognitive arousal. PSAS-C scores ≥ 20 suggest clinically significant insomnia and affective symptoms (Puzino et al., 2019).

The Perseverative Thinking Questionnaire (PTQ) is a 15-item transdiagnostic measure of repetitive thinking (Ehring et al., 2011), which was first administered after screening but before treatment. The valence of many PTQ items is neutral (e.g., My thoughts repeat themselves and Thoughts just pop into my mind), although the valence of some items is negative (e.g., I think about many problems without solving them). Higher scores on the PTQ indicate greater tendency to engage in transdiagnostic repetitive thinking.

The Ruminative Response Scale (RRS) is a 22-item self-report measure of ruminative thought that originated in depression research (Treynor et al., 2003). The RRS was first administered after screening but before treatment in this RCT. Although the measure is often used as a unitary construct, three subscales have been identified: focusing on depressive symptoms (e.g., think about how sad you feel), brooding (e.g., think ‘what am I doing to deserve this?’), and reflective pondering about the causes of negative affect (e.g., go someplace alone to think about your feelings). The depressive symptom-focused subscale is highly collinear with depression, whereas the brooding and pondering subscales are less confounded (Treynor et al., 2003). In the present study, we omitted the depression-focused subscale and only administered items from the brooding and pondering scales. Higher scores on the RRS indicate higher rumination.

Sociodemographics, health information, and perinatal factors.

At screening, patients provided standard sociodemographic and health information including age, race, and parity. Patients reported whether they snore (Yes/No) and estimated the timing of their insomnia onset, from which we derived whether insomnia onset before or during pregnancy. We collected data on number of sessions completed to assess adherence. Body mass index (BMI; kg/m2) was derived from electronic medical records from visit notes closest to the screening date. After childbirth, new moms were asked to report on how well their infant was sleeping on a 4-point Likert-type scale, which ranged from ‘1=Very well, rarely wakes in the middle of the night and easy to get to sleep’ and ‘2=Somewhat well’ to ‘3=Somewhat poorly’ and ‘4=Very poorly, waking up frequently, hard to get to sleep.’

Patient feedback inquiries

At the completion of study participation, CBTI patients were invited to provide treatment feedback. Patient feedback was collected via 16 questions constructed by the study team. Thirteen of these questions were accompanied by Likert-type response scales, whereas the other 3 were open-ended. We sought feedback on online appeal, feasibility of in-person CBTI treatment, ease/difficulty attending in-person health appointments during late pregnancy, how helpful/unhelpful CBTI was for sleep during and after pregnancy, satisfaction/dissatisfaction with CBTI, how easy/difficult CBTI was to complete, how engaging digital CBTI was, whether digital CBTI was sufficient to alleviate their sleep problems, CBTI effects on nighttime worries and stressful thoughts, and comparisons between in-person vs digital treatment and in-person vs telemedicine treatment. Please note that each feedback inquiry and associated response scale is listed verbatim in the Results section during exploration of patient feedback; they are not reported here to avoid redundancy within the manuscript.

Analysis plan

All analyses were performed in SPSS version 25 (IBM Corp) with alpha set at .05. Important to highlight here is that we previously reported RCT treatment results for ISI, PSQI, EPDS, and PSAS-C study outcomes (Kalmbach, Cheng, O’Brien, et al., 2020). In the present report, we report RCT results for two additional cognitive arousal study outcomes: perseverative thinking (PTQ) and depressive rumination (RRS). As these data are not central to the present study’s goals but were rather conducted for context and completeness, these results are reported in Supplementary Table 1. Specifically, we compared study outcomes at pretreatment, posttreatment, and postnatal follow-up using independent samples t-tests to determine whether CBTI patients and controls differed on perseverative thinking or depressive rumination at any point during the trial. Additional information on the control group can be found in the supplementary materials.

To achieve the goals of the present study, we first reported sociodemographics and study outcome levels at each assessment for the CBTI group. A primary focus of this study was to identify prenatal and postnatal patient characteristics associated with non-response to CBTI. To achieve this goal, we used independent samples t-tests (continuous variables) and chi-square analyses (binary variables) to compare responders and non-responders on sociodemographic information, treatment adherence, and study outcome values at pretreatment, posttreatment, and postnatal follow-up. Ancillary objectives were to identify patient characteristics associated short sleep duration in early parenting (≤6 hrs/night). Thus, we repeated this process for comparing postpartum participants with normal and short sleep duration. To elucidate the potential role of cognitive arousal symptoms in treatment response, we conducted posthoc bivariate correlations to test whether changes in insomnia and depression symptoms (between pre and posttreatment) were correlated with changes in cognitive arousal (nocturnal cognitive arousal, perseverative thinking, depressive rumination).

Our second study goal was to characterize patient feedback to CBTI, which involved a descriptive analytic approach. Because not all CBTI patients provided treatment feedback, we first used independent samples t-tests and chi-square analyses to compare patients who did and did not provide feedback to identify any differences. Next, we reported frequency distributions for all 13 patient feedback inquiries that were accompanied by Likert-type response scales. This descriptive analysis was followed by exploration of open-ended responses to feedback inquiries on digital treatment delivery, and positive and negative impressions of digital CBTI. Themes mentioned by multiple patients are reported in the results section. In addition, open-ended responses were entered into word cloud generators to create visual representations of patient feedback.

Results

Sample characteristics

A total of 46 women were randomized to CBTI in our RCT (see Table 1 for sociodemographics and pretreatment values of study outcomes). Patients mostly identified as non-Hispanic white or non-Hispanic black. Regarding the number of CBTI sessions completed, the median and mode were 6.0 (i.e., completed all sessions), whereas the mean was 4.83 sessions. As designed, pretreatment ISI and PSQI mean scores reflected clinically significant insomnia symptoms and poor sleep quality, and average nightly sleep duration was relatively short. Pretreatment mean EPDS scores were slightly below the clinical cutoff for probable depression (8.45±4.47) and mean PSAS-C scores were elevated (22.34±6.54) indicating high nocturnal cognitive arousal.

Table 1.

Comparing sociodemographics and study outcomes at pretreatment, posttreatment, and 6-week postnatal follow-up between CBTI responders and non-responders.

	Full CBTI Group	Responders	Non-Responders
Sample size	46	17	29
Age (M±SD)	28.91 ± 4.25	29.06±4.55	28.83±4.15	t(44)=.18, p=.86
Poverty (n;%)	7/46; 15.2%	4; 23.5%	3; 10.3%	χ2=1.44, p=.23
BMI	9/44; 20.5%	29.48±4.91	30.72±4.98	t(42)=−.80, p=.43
Multiparous (n;%)	17/46; 37.0%	6; 35.3%	11; 37.9%	χ2=.03, p=.86
Snore (n;%)	19/46; 41.3%	8; 47.1%	11; 37.9%	χ2=.37, p=.54
Gestational insomnia (n;%)	9/46; 19.6%	3; 17.6%	6; 20.7%	χ2=.06, p=.80
Race (n;%)				χ2=2.83, p=.73
White	24; 52.2%	11; 64.7%	13; 44.8%
Black	15; 32.6%	4; 23.5%	11; 37.9%
Asian	3; 6.5%	1; 5.9%	2; 6.9%
Middle Eastern or Arabic	2; 4.3%	1; 5.9%	1; 3.4%
Hispanic or Latino	1; 2.2%	0; 0.0%	1; 3.4%
Multiracial	1; 2.2%	0; 0.0%	1; 3.4%
Sessions completed (M±SD)	4.83 ± 1.88	5.24±1.68	4.59±1.97	t(44)=1.14, p=.26
Median	6.0	6.0	6.0
Mode	6.0	6.0	6.0
Study Outcomes
ISI
Pretreatment	14.91±3.55	16.24±3.93	14.14±3.11	t(44)=2.00, p=.05
Posttreatment	10.00±5.66	5.23±3.68	12.79±4.69	t(44)=−5.69, p<.001
Follow-up	8.96±5.46	7.41±4.68	9.86±5.76	t(44)=−1.49, p=.14
PSQI
Pretreatment	9.59±2.61	9.53±2.48	9.62±2.73	t(44)=−.11, p=.91
Posttreatment	6.61±2.99	4.65±1.87	7.76±2.95	t(44)=−3.91, p<.001
Follow-up	7.00±3.22	6.12±2.87	7.52±3.34	t(44)=−1.44, p=.16
Sleep Duration (in hrs)
Pretreatment	6.32±1.26	6.65±1.07	6.12±1.34	t(44)=1.38, p=.18
Posttreatment	6.85±1.17	7.44±1.00	6.50±1.14	t(44)=2.82, p=.007
Follow-up	6.10±1.53	6.38±1.69	5.93±1.43	t(44)=.97, p=.34
EPDS
Pretreatment	7.46±4.14	7.47±4.03	7.45±4.03	t(44)=.02, p=.99
Posttreatment	5.87±4.28	4.24±3.95	6.83±4.24	t(44)=−2.05, p<.05
Follow-up	4.76±4.07	4.53±4.06	4.90±4.14	t(44)=−.29, p=.77
PSAS-C
Pretreatment	22.02±6.09	23.06±5.57	21.41±6.39	t(44)=.88, p=.38
Posttreatment	15.70±6.16	11.42±2.62	18.21±6.27	t(44)=−4.24, p<.001
Follow-up	14.35±5.96	11.82±3.71	15.83±6.74	t(44)=−2.30, p=.03
PTQ
Pretreatment	21.74±11.90	20.53±11.71	22.45±12.16	t(44)=−.52, p=.60
Posttreatment	12.83±12.65	5.18±6.06	17.31±13.41	t(44)=−3.51, p=.001
Follow-up	13.40±9.57	6.75±10.63	11.18±14.68	t(44)=−1.06, p=.30
RRS
Pretreatment	18.91±6.82	18.76±6.15	19.00±7.29	t(44)=−.11, p=.91
Posttreatment	14.85±6.23	12.65±3.95	16.14±6.98	t(44)=−1.89, p=.07
Follow-up	13.55±6.38	12.13±3.83	14.36±7.40	t(44)=−1.12, p=.27

CBTI = cognitive behavioral therapy for insomnia. Responders = CBTI patients who reported decrease in Insomnia Severity Index score of ≥6 or more points between pre and posttreatment assessments. Non-Responders = CBTI patients who reported reductions of ≤ 5 points on the Insomnia Severity Index. M = mean. SD = standard deviation. n = number of patients. t = t-statistic for independent samples t-test. χ² = chi-square test value. p = significance value. Poverty ≤ $20,000 annual household income. BMI = body mass index. ISI = Insomnia Severity Index. PSQI = Pittsburgh Sleep Quality Index. Sleep duration assessed via PSQI item 4a. EPDS = Edinburgh Postnatal Depression Scale. PSAS-C = Presleep Arousal Scale, Cognitive factor. PTQ = Perseverative Thinking Questionnaire. RRS = Ruminative Response Scale.

Identifying factors associated with poor CBTI outcomes.

CBTI non-response (reduction of < 6 points on the ISI between pre and posttreatment).

CBTI responders and non-responders did not differ on sociodemographic factors, sessions completed, nor pretreatment clinical symptom levels. However, several posttreatment study outcomes differed between responders and non-responders (Table 1). Aside from having poorer sleep quality (Cohen’s d=1.26) and shorter sleep duration (56 fewer minutes per night) after treatment, non-responders reported moderately more severe depressive symptoms (Cohen’s d=.63) and substantially more nocturnal cognitive arousal (Cohen’s d=1.41) and perseverative thinking (Cohen’s d=1.17) than responders.

Short sleep in early postpartum.

Twenty-five of 44 women (54.3%) who provided postpartum data estimated sleeping ≤ 6 hrs per night. Postpartum women with short sleep duration reported poorer infant sleep quality than postpartum women with normal sleep duration, which was a medium-large effect (Cohen’s d=.69; see Supplementary Table 2 for all comparisons). Indeed, 20% of postpartum women with short sleep duration perceived their infant’s sleep as poor, whereas 0% of postpartum women who slept longer than 6 hours per night rated their infant’s sleep as poor. Some data suggested that postpartum women with short sleep may experience elevated cognitive arousal during pregnancy and postpartum, but the pattern was not consistent across indices in this study, and thus should be interpreted with some caution (see PSAS-C, PTQ, and RRS data in Supplementary Table 2).

Associations among treatment-related changes in insomnia, depression, and cognitive arousal.

We then conducted posthoc bivariate correlations to explore whether changes in insomnia and depression were associated with changes in cognitive arousal. Analyses clearly revealed that acute and longer-term changes in insomnia and depression were consistently associated with changes in nocturnal cognitive arousal and perseverative thinking such that patients whose insomnia and depression symptom levels decreased also reported reductions in nocturnal cognitive arousal and perseverative thinking (Table 2). Reductions in depressive rumination were associated with alleviation of depressive symptoms, whereas correlations between changes in depressive rumination and insomnia only trended toward significance at posttreatment (p=.08) and postnatal follow-up (p=.07).

Table 2.

Bivariate correlations between (Top) changes in insomnia symptoms and changes in depressive symptoms and indices of cognitive arousal, and (Bottom) changes in depressive symptoms and changes in indices of cognitive arousal.

	Pre to Posttreatment	Pre to Follow-up
	Δ ISI
Δ EPDS	.32*	.42**
Δ PSAS-C	.64***	.59***
Δ PTQ	.40**	.41**
Δ RRS	.26?	.27?
	Δ EPDS
Δ PSAS-C	.31*	.43**
Δ PTQ	.50***	.64**
Δ RRS	.42**	.62***

Note: Δ denotes changes in scores from pretreatment to posttreatment (left column), or pretreatment to postnatal follow-up (right column). ISI = Insomnia Severity Index. EPDS = Edinburgh Postnatal Depression Scale. PSAS-C = Presleep Arousal Scale, Cognitive factor. PTQ = Perseverative Thinking Questionnaire. RRS = Ruminative Response Scale.

^*p<.05.

^**p<.01.

^***p<.001.

^?p<.10.

Characterizing CBTI feedback

A total of 26 out of 46 CBTI patients (56.5%) provided feedback after study completion (see Table 3 for group comparisons). Patients who provided feedback, relative to those who did not, completed more sessions, yet reported more residual sleep disturbance after treatment on the PSQI. Even so, these groups did not differ on insomnia symptom severity at any time within the study treatment period. The groups did not differ on sociodemographics.

Table 3.

Sample demographics and characteristics by feedback group.

	Full CBTI Group	Feedback	No Feedback
Sample size	46	26	20
Age (M±SD)	28.91 ± 4.25	29.85 ± 4.56	27.70 ± 3.57	t(44)=1.73, p=.09
Poverty (n;%)	7/46; 15.2%	4; 15.4%	3; 15.0%	χ2=.00, p=.97
BMI	9/44; 20.5%	29.96±5.00	30.68±4.95	t(42)=−.48, p=.64
Multiparous (n;%)	17/46; 37.0%	7; 26.9%	10; 50.0%	χ2=2.58, p=.11
Snore (n;%)	19/46; 41.3%	13; 50.0%	6; 30.0%	χ2=1.87, p=.17
Gestational insomnia (n;%)	9/46; 19.6%	3; 11.5%	6; 30.0%	χ2=2.45, p=.12
Race (n;%)				χ2=5.21, p=.39
White	24; 52.2%	15; 57.7%	9; 45.0%
Black	15; 32.6%	8; 30.8%	7; 35.0%
Asian	3; 6.5%	1; 3.8%	2; 10.0%
Middle Eastern or Arabic	2; 4.3%	2; 7.7%	0; 0.0%
Hispanic or Latino	1; 2.2%	0; 0.0%	1; 5.0%
Multiracial	1; 2.2%	0; 0.0%	1; 5.0%
Sessions completed (M±SD)	4.83 ± 1.88	5.62 ±.94	3.80±2.28	t(44)=3.68, p=.001
Median	6.0	6.0	6.0
Mode	6.0	6.0	6.0
Study Outcomes
ISI
Pretreatment	14.91±3.55	14.65±3.61	15.25±3.52	t(44)=−.56, p=.58
Posttreatment	10.00±5.66	10.27±6.05	9.65±5.25	t(44)=.36, p=.72
Follow-up	8.96±5.46	9.42±4.94	8.35±6.15	t(44)=.66, p=.52
PSQI
Pretreatment	9.59±2.61	9.77±2.70	9.35±2.54	t(44)=.54, p=.60
Posttreatment	6.61±2.99	7.54±3.23	5.40±2.19	t(44)=2.55, p=.02
Follow-up	7.00±3.22	7.69±3.31	6.10±2.94	t(44)=1.70, p<.10
Sleep Duration (in hrs)
Pretreatment	6.32±1.26	6.21±1.37	6.45±1.12	t(44)=−.63, p=.53
Posttreatment	6.85±1.17	6.62±1.13	7.15±1.19	t(44)=−1.56, p=.13
Follow-up	6.10±1.53	5.75±1.56	6.55±1.40	t(44)=−1.80, p=.08
EPDS
Pretreatment	7.46±4.14	7.50±3.87	7.40±4.57	t(44)=.08, p=.94
Posttreatment	5.87±4.28	5.58±4.35	6.25±4.28	t(44)=−.52, p=.60
Follow-up	4.76±4.07	4.88±4.07	4.60±4.17	t(44)=.23, p=.82
PSAS-C
Pretreatment	22.02±6.09	21.65±6.28	22.50±5.96	t(44)=.46, p=.65
Posttreatment	15.70±6.16	16.42±6.42	14.75±5.82	t(44)=.91, p=.37
Follow-up	14.35±5.96	15.27±5.24	13.15±6.74	t(44)=1.20, p=.24

CBTI = cognitive behavioral therapy for insomnia. Responders = CBTI patients who reported decrease in Insomnia Severity Index score of ≥6 or more points between pre and posttreatment assessments. Non-Responders = CBTI patients who reported reductions of ≤ 5 points on the Insomnia Severity Index. M = mean. SD = standard deviation. n = number of patients. t = t-statistic for independent samples t-test. χ² = chi-square test value. p = significance value. Poverty ≤ $20,000 annual household income. BMI = body mass index. ISI = Insomnia Severity Index. PSQI = Pittsburgh Sleep Quality Index. Sleep duration assessed via PSQI item 4a. EPDS = Edinburgh Postnatal Depression Scale. PSAS-C = Presleep Arousal Scale, cognitive factor. PTQ = Perseverative Thinking Questionnaire. RRS = Ruminative Response Scale.

We next examined patient impressions of digital delivery of treatment (see Table 4 for frequency distributions). Nearly 1 in 5 women reported that attending health appointments was more difficult in late pregnancy as compared with pre-pregnancy. All patients stated that the online nature of the intervention had a medium or large influence on their decision to participate. Notably, 65.4% of patients indicated that they would likely not have participated if required to attend CBTI in-person.

Table 4.

Distributions of patient feedback on digital CBTI for the treatment of perinatal insomnia.

	None	Small	Medium	Large
How much of an influence did the internet-based nature of the study have on your decision to participate and engage in treatment?	0%	0%	23.1%	76.9%
	Absolutely Not	Probably Not	Maybe	Absolutely Yes
Would you have been able to complete your treatment if you were required to visit our sleep clinic in person for 6 sessions?	30.8%	34.6%	19.2%	15.4%
	Easier than pre- pregnancy	Not at all more difficult	Somewhat more difficult	Much more difficult
During 3rd trimester, how difficult was it to attend health appointments?	19.2%	61.5%	15.4%	3.8%
	Very unhelpful	Somewhat unhelpful	Neutral	Somewhat helpful	Very helpful
How helpful was digital CBTI for your sleep DURING pregnancy?	7.7%	3.8%	3.8%	46.2%	38.5%
How helpful was your treatment for your sleep AFTER delivery?	19.2%	19.2%	15.4%	38.5%	7.7%
	Very dissatisfied	Dissatisfied	Neutral	Satisfied	Very satisfied
How satisfied are you with your sleep treatment?	0%	7.7%	11.5%	46.2%	34.6%
	Very difficult	Somewhat Difficult	Neither easy nor difficult	Somewhat easy	Very easy
How easy was your sleep treatment to complete?	0%	7.7%	3.8%	30.8%	57.7%
	Not at all	Minimally	Somewhat	Moderately	Very
How engaging was treatment?	3.8%	3.8%	23.1%	26.9%	42.3%
	Very insufficient	Insufficient	Neutral	Sufficient	Very sufficient
How sufficient was your treatment for your sleep problem?	3.8%	7.7%	11.5%	57.7%	19.2%
	Not at all	A little	Somewhat	Moderately	Very much
Did treatment improve your nighttime worries or stressful thoughts?	7.7%	11.5%	11.5%	30.8%	38.5%
Has your sleep issue resolved as a result of your sleep treatment?	0%	15.4%	26.9%	42.3%	15.4%
	In-person	Digital online		In-person	Telemedicine online video
In the future, if you were offered treatment for sleep, which would you prefer?	15.4%	84.6%		19.2%	80.8%

Regarding treatment appraisal (see Table 4), 80.8% of patients reported being satisfied overall with CBTI and 76.9% described it as a sufficient treatment for their insomnia. Even so, only 57.7% of patients indicated that their sleep issues were resolved (and < 40% responded per ISI). Most patients (84.7%) described digital CBTI as helpful for sleep in pregnancy, but less than half of patients described CBTI as helpful for postpartum sleep. Most patients described digital CBTI as easy (88.5%) and engaging (69.2%), and indicated that CBTI helped reduce nocturnal cognitive arousal and perseverative thinking (69.3%) despite a lack of treatment results in the RCT. When given the choice between in-person sleep treatment vs digital online or telemedicine online video, most patients (>80%) preferred online or telemedicine treatment over in-person treatment.

Patient open-ended, qualitative feedback.

Our first open-ended inquiry asked patients to express their thoughts comparing digital and in-person delivery of treatment (see Supplementary Figure 1 for word cloud). The most common themes described online treatment as ‘convenient’ (n=14), ‘easy’ (n=8), and ‘flexible’ (n=6). Several women mentioned time, travel, and work-related conflicts (n=11), and responsibilities with other children at home and not needing childcare for online treatment (n=8).

Next, we asked patients to indicate the most helpful aspects of digital CBTI (see Supplementary Figure 2 for word cloud). The most commonly identified helpful aspects of digital CBTI centered on sleep hygiene tips and sleep education (n=8), reduction of sleep-interfering nighttime thoughts (n=6), self-monitoring with sleep diaries (n=5), the virtual therapist (n=3), sleep restriction (n=3), stimulus control (n=3), and mindfulness meditation (n=2).

Patients also reported what they considered to be limitations CBTI or areas in need of improvement (see Supplementary Figure 3 for word cloud). The most commonly cited drawbacks were that digital CBTI was not tailored for pregnancy (n=6) nor postpartum sleep challenges (n=6). The online program was described as too strict and inflexible regarding sleep restriction and sleep schedules during pregnancy and early postpartum (n=7).

Discussion

In a sample of 46 women with insomnia who completed digital CBTI during pregnancy, we identified patient characteristics associated with treatment non-response and postpartum short sleep, and described patient feedback on treatment. Women who did not respond to CBTI, relative to responders, reported higher levels of cognitive arousal and depressive symptoms after (but not before) treatment. In the early newborn phase, maternal short sleep was linked to poor infant sleep quality. Patients appreciated the ease, convenience, and flexibility of digital treatment, and described CBTI as helpful for sleep and sleep-interfering thoughts. Even so, patients also indicated that CBTI needs to be tailored to meet the challenges unique to pregnancy and early postpartum. Carefully designed enhancements and modifications to CBTI will be critical to maximize treatment efficacy and patient uptake. Overall, these study findings may help guide modifications to better tailor insomnia therapy for the perinatal experience. While many results were consistent with data from RCTs in other patient populations, other study observations were without precedent, such as our patient feedback and the association of posttreatment short sleep with CBTI non-response. Replication of study findings in larger samples will be important for the refinement of insomnia therapy for perinatal insomnia.

Refractory cognitive arousal in CBTI non-response.

Although standard CBTI reduces insomnia-focused rumination (Ballesio et al., 2018; Sunnhed & Jansson-Fröjmark, 2014), RCT data offer weak support for CBTI improving non-sleep-specific forms of cognitive arousal (e.g., presleep cognitive arousal, worry, depressive, stress-focused rumination). Indeed, multiple studies show minimal or no significant CBTI effect on cognitive arousal relative to attention control (Cheng et al., 2020; Kalmbach, Cheng, Arnedt, et al., 2019; Kalmbach, Cheng, O’Brien, et al., 2020). While these findings may seem initially divergent, it is possible that CBTI’s well-documented effects on sleep quality drives reductions in insomnia-focused rumination. In other words, if you sleep well, you are not likely to ruminate on sleep problems that you no longer have. But this does not mean that your typical approach to emotion regulation has changed; i.e., you may still be a ruminator or worrier, just no longer ruminating on sleep problems.

Standard CBTI’s minimal effect on cognitive arousal is an important limitation for several reasons. Cognitive arousal is a central feature of insomnia etiology and chronicity (Harvey, 2005; Ong et al., 2012). When CBTI produces even small treatment effects on cognitive arousal, the reduction of cognitive arousal is a key mechanism in the alleviation of insomnia and depression, as well as in depression prevention (Cheng et al., 2020). Moreover, refractory cognitive arousal is associated with blunted response to CBTI, such that patients whose presleep cognitive arousal and perseverative thinking do not decrease with treatment are less likely to adequately respond (Cheng et al., 2020; Kalmbach, Cheng, Arnedt, et al., 2019). Even among patients who initially remit with insomnia therapy, residual cognitive arousal symptoms after treatment increase risk for relapse within 1 year (Ong et al., 2009).

In the present RCT, digital CBTI did not improve nocturnal cognitive arousal, depressive rumination, or perseverative thinking relative to control (see (Kalmbach, Cheng, O’Brien, et al., 2020) and Supplementary Table 1). Problematically, high levels of cognitive arousal after treatment were associated with non-response to CBTI for pregnant women and postnatal short sleep, thereby further supporting refractory cognitive arousal as an obstacle to adequate treatment response. This shortcoming of CBTI is dangerous for several reasons. If reducing cognitive arousal is a mechanism by which CBTI alleviates insomnia and depression, then standard CBTI is ill-equipped to maximally improve sleep and mood in perinatal women. Further, high cognitive arousal is strongly linked to prenatal depression and suicidal thoughts (DeJong et al., 2016; Kalmbach, Cheng, Ong, et al., 2020; O'Mahen et al., 2010). Insomnia therapies for perinatal women need to successfully target cognitive arousal to maximize patient outcomes and minimize risk for suicidality.

Patient feedback & insights for tailoring CBTI for perinatal women

As expected in a self-selecting sample in a digital treatment study, most patients were attracted to the ease and flexibility of digital CBTI. Many women reported that digital intervention allowed them to navigate their treatment around scheduling issues related to work hours, childcare, and travel difficulties. While many women were satisfied with the treatment and found digital CBTI helpful, we observed no alleviation of cognitive arousal, which patients identified as an important treatment component. Crucially, patients clearly indicated that insomnia therapy would be more helpful if tailored to meet the changing needs of women as they progress through pregnancy and then postpartum.

Enhancing treatment effects on cognitive arousal.

Many pregnant women, particularly those with insomnia, endorse pathogenic levels of cognitive arousal during pregnancy. Not only is cognitive hyperarousal linked to depression, suicidality, and impaired mother-to-infant bonding in peripartum, but ruminating on pregnancy, childbirth, and fetal/infant health concerns may a play a uniquely important role in maternal sleep disturbance and diminished wellbeing (Kalmbach, Cheng, Ong, et al., 2020; Martini et al., 2016; Müller et al., 2013). Enhancing insomnia therapy effects on cognitive arousal—including specific attention to ruminative thoughts on pregnancy, childbirth, and fetal/infant health—may not only be important to optimizing sleep outcomes, but may also improve overall maternal wellbeing.

Despite our RCT data showing no treatment effects of CBTI on cognitive arousal, over 2/3 of CBTI patients who provided treatment feedback reported that digital CBTI helped reduce sleep-interfering thinking. This feedback indicates that cognitive arousal is an important therapeutic target according to perinatal insomnia patients. Potential strategies to better alleviate cognitive arousal may include increasing from a traditional 1-2 sessions to 4-5 of cognitive therapy in CBTI. Alternatively, augmenting CBTI with rumination-focused content may beneficial. Rumination-focused cognitive-behavioral therapy (RF-CBT) was developed to enhance rumination outcomes in depressed patients (Watkins et al., 2011). Combining components of RF-CBT with CBTI could potentially enhance acute and long-term insomnia outcomes and improve depression outcomes by reducing ruminative coping and presleep cognitive arousal.

Along these lines, cognitive therapy for insomnia (Harvey, 2005) may better alleviate cognitive arousal and ruminative thinking than standard CBTI, and its cognitive focus may be more palatable for women who struggle with the behavioral components of CBTI. Pivoting away from cognition-focused strategies, mindfulness-based and emotion regulation interventions are highly successful at reducing cognitive arousal in the forms of rumination and worry (Deyo et al., 2009; Mennin et al., 2018). Indeed, mindfulness-based therapy for insomnia (MBTI), which places behavioral sleep strategies within a mindfulness intervention framework, substantially improves both insomnia and cognitive arousal symptoms in the broader insomnia population (Ong et al., 2014).

Regardless of the therapeutic approach, these modifications to enhance insomnia therapy effects on cognitive arousal will need to be first explored and tested with therapists who can deliver highly personalized arousal-reducing strategies. Importantly, skilled clinicians can help women process concerns related to pregnancy, childbirth, and fetal/infant health, which may be especially therapeutic in this patient population. Once these enhancements are supported and refined, they may be able to inform digital treatments.

Tailoring insomnia therapy for peripartum.

Several unique factors can trigger and perpetuate sleep problems during pregnancy and early parenting, including pregnancy/maternal-specific worries, physical discomfort and pain, fetal activity, nocturnal infant feedings, poor infant sleep, difficult infant temperament, poor family support (Bayer et al., 2007; Hunter et al., 2009; Krystal, 2003; Martini et al., 2016; Meltzer & Mindell, 2007; Reichner, 2015; Thomas & Foreman, 2005). Sleep treatments for perinatal women may benefit from providing education and normalization of these experiences, in addition to behavioral and cognitive strategies to manage these sleep-interfering challenges. Indeed, other teams have included psychoeducation on maternal and infant sleep in CBTI protocols for pregnant and postpartum women, which may have enhanced treatment outcomes (Manber et al., 2019; Swanson et al., 2013). Going forward, it will be important to further investigate, refine, and eventually standardize these recommendations for this patient population.

Modifying behavioral sleep strategies.

Pregnant and postpartum women reported difficulty adhering to rigid behavioral sleep strategies; namely, sleep restriction and maintaining regular sleep schedules (example quotes: “restricted sleep window was not helpful when pregnant,” “going to bed at a specific time and waking up, it was not helpful when you have kids waking up”). These adherence difficulties necessitate the development of behavioral sleep strategies that can improve sleep efficiency and stabilize sleep-wake patterns (like sleep restrictions and prescribed sleep schedules), but also allow for flexibility to compensate for sleep deprivation (e.g., due to nighttime feedings or dysregulated infant sleep) and childcare.

CBTI can be modified in several ways to better accommodate prenatal and postpartum sleep challenges. In the present study, the only behavioral modification to CBTI was to ensure that time-in-bed was never restricted below 6 hrs; this modification alone is inadequate. In the published in-person CBTI RCT for prenatal insomnia, the authors included this modification of never restricting time-in-bed to < 6 hrs, but also added an extra 30 minutes to time-in-bed (Manber et al., 2019). In a CBTI clinical trial in postpartum women, Swanson et al. provided patients with bed- and wake-time windows (30-60-min) to accommodate variable infant sleep patterns, which was a decision based on focus group feedback (Swanson et al., 2013). Notably, allowing flexibility in bed/wake-times is not likely to reduce treatment efficacy based on recent data suggesting that rigid sleep schedules are not necessary for insomnia therapy success (Shaffer et al., 2020).

Owing in part to unconsolidated newborn infant sleep, maternal sleep is often short and highly disrupted in early postpartum, which can increase daytime sleepiness and fatigue (Insana & Montgomery-Downs, 2010). Behavioral sleep strategies to offset daytime sleepiness may involve incorporating napping into CBTI for perinatal insomnia. In Swanson’s prior CBTI trial in postpartum women, patients were permitted short naps (≤ 30 minutes), which patients identified as helpful. Timing naps in a CBTI regiment must be carefully considered. For non-perinatal patients, guidance for timing the nap has been 7-9 hours after wake time, so that napping coincides with the natural dipping of the circadian alerting system (Manber et al., 2014). However, women in early postpartum are often limited in the opportunity to nap based on the infant’s napping and feeding needs, and the availability of caregiving assistance from a partner or other individuals. A short nap in the morning or afternoon may be sufficient to offset daytime sleepiness without disrupting nocturnal sleep, whereas evening napping runs the risk of reducing sleep drive in close proximity to bedtime, which may disrupt nocturnal sleep.

Insomnia therapy to address postpartum sleep challenges.

Patients reported that CBTI strategies that were effective during pregnancy were less helpful during the early postnatal period. Given that poor and insufficient maternal sleep after childbirth is associated with neurobehavioral impairment and, thus, could be dangerous for mother and child (Insana et al., 2011; Insana et al., 2013), discontinuing sleep treatment during pregnancy (even if successful) may premature. Most women who received CBTI in pregnancy reported continued sleep disturbance after delivery, although the nature of symptoms sometimes changed (e.g., reduced sleep duration, infant awakenings impacting maternal sleep). Thus, continued monitoring of maternal sleep should be encouraged during postnatal care. Unsurprisingly, over half of postpartum women who received CBTI reported sleeping < 6 hrs per night during early postpartum. While this rate is high, 91% of controls in our RCT who received sleep hygiene education information during pregnancy also reported short sleep in postpartum (Kalmbach, Cheng, O’Brien, et al., 2020), thereby suggesting that prenatal insomnia treatment offers protection against postpartum short sleep.

Along these lines, infant sleep problems can have negative consequences on maternal sleep and mood (Meltzer & Montgomery-Downs, 2011). Indeed, we found poor infant sleep quality was strongly associated with short sleep duration in the newborn period. Prior research shows that, even when partners are available, mothers report greater time awake at night, relative to father, tending to infant feedings, general infant care, and infant changing (Insana et al., 2014). Encouraging more equitable nocturnal caregiving when available would likely improve maternal sleep, which is an important augmentation utilized previously in a CBTI for postpartum sleep trial (Swanson et al., 2013). Rather than simply suggest to the patient that she devise a schedule with her partner, clinicians should collaborate with the patient (and partner) to construct a concrete shift schedule, which may help promote patient uptake and follow-through.

As sleep begins to consolidate by 6 months for most infants (Meltzer & Montgomery-Downs, 2011), postpartum maternal sleep treatment should include teaching mothers behavioral strategies to promote healthy infant sleep, such as sleep training, when mother and child are ready. Indeed, clinical trials show that improving infant sleep with behavioral sleep strategies also improves maternal sleep (Gradisar et al., 2016; Hiscock et al., 2007; Wolfson et al., 1992).

Limitations

The present study should be interpreted in light of certain limitations. Primary analyses in the present study compared treatment responders vs non-responders. These groups were derived from ISI scores that have not been specifically validated in a pregnant or postpartum patient population. Other self-report metrics of treatment response should be considered in future investigations of insomnia treatment in perinatal women. Additionally, all outcomes were self-reported and limited to maternal perceptions. This limitation may be most relevant to ratings of infant sleep. It is unclear whether, or to what extent, maternal depression or sleep problems may influence maternal ratings of infant sleep. Moreover, infant and maternal sleep changes drastically in early postpartum. Future studies should consider multiple assessments of infant and maternal sleep across a longer period of postpartum to provide more nuanced characterization of important parameters, such as maternal short sleep.

Another primary limitation concerns only collecting patient feedback from 26 of the 46 digital CBTI patients, which may limit some of the generalizability of the patient feedback. As the feedback interview was uncompensated and administered at the end of participation, it is possible that only the most self-motivated postpartum women were interested in providing treatment feedback (which may explain why feedback providers completed more sessions than patients who did not provide feedback). Notably, feedback providers also reported more residual sleep disturbance symptoms after treatment, so it is possible that residual sleep symptoms motivated these patients to offer feedback on how the program can be improved. In any case, it is unclear how this self-selecting subsample’s feedback may differ from those who were less inclined to provide feedback. Even so, it is worth emphasizing that patients who did and did not provide feedback largely did not differ from one another on key study variables.

Patient opinions on the ease and preferability of digital treatment is not generalizable to all pregnant women with insomnia, because the present study’s sample self-selected into a digital treatment study. Thus, it would not be accurate to characterize most pregnant women as preferring digital treatment. Rather, we can simply conclude that a subset of pregnant women prefers digital (and even real-time video) healthcare over in-person treatment for the reasons outlined above. Future studies are needed to better characterize patient preferences in the broader perinatal population regarding preferences for in-person, real-time video/telephone, and fully automated digital healthcare options. This information will be key for identifying treatment delivery methods that maximize patient access and likelihood of uptake.

Our limited follow-up assessment represents another important limitation of the study. Postpartum women with short sleep endorsed poorer infant sleep quality. As infant sleep continues to stabilize after the newborn period, some postpartum women will likely obtain more sleep, whereas some will continue to sleep poorly. Future studies that can identify women at risk for insomnia and/or short sleep even after infant sleep stabilizes are needed to identify therapeutic targets to help these women obtain adequate good quality sleep.

Finally, our study consisted of multiple comparisons, which increases risk for type I error. Due to small sample size, we did not adjust our alpha, which would have resulted in inadequate power. Even so, our data need to be interpreted within the context of the extant literature and across measures within the study. For instance, prior RCTs by our team and others have shown that refractory cognitive arousal is linked to poor response to insomnia therapy (Cheng et al., 2020; Kalmbach, Cheng, Arnedt, et al., 2019; Ong et al., 2009), which was a consistent finding across multiple measures in our study. Further, we have previously shown that nocturnal cognitive arousal is associated with objective short sleep (Kalmbach, Buysse, et al., 2020). Novel findings, such as the associations of maternal short sleep with CBTI non-response, should be replicated and further explored in future studies with larger samples.

Conclusions

Data from this RCT and prior trials suggest that standard CBTI improves sleep in pregnancy and postpartum, but that optimizing treatment outcomes in this patient population requires creative modifications to meet the unique and changing needs of women as they progress through pregnancy and transition to early parenting. Critical areas of improvement should focus on providing psychoeducation on maternal and infant sleep, reducing cognitive arousal particularly at night, modifying behavioral sleep strategies to allow for more patient flexibility and sleep opportunity (which may include restricted and appropriately-timed napping), reducing maternal burden for nighttime infant feedings and wakings, and teaching mothers behavioral strategies to improve and stabilize infant sleep when the child is ready. Initial efforts to optimize insomnia therapy in perinatal patients will be best suited for specialist-driven care, given the limits of treatment personalization inherent to digital interventions. At this stage, telemedicine CBTI for perinatal insomnia offers a strong balance between increasing treatment access and clinician-led treatment personalization. Once perinatal insomnia therapy is refined, the development of fully automated digital insomnia therapy for perinatal insomnia can be designed, which can further increase treatment access.

Data availability statement:

The data that support the findings of this study are available from the corresponding author, DAK, upon reasonable request.