Examining changes in presenteeism and clinical symptoms in a workforce mental health benefits program

Abstract

This study analyzed retrospective data from clients who utilized a workforce mental health benefit (WMHB) to understand how changes in anxiety and depression symptoms may predict changes in presenteeism (n = 704). Results indicated that presenteeism improved during WMHB care episodes and the effect varied based on the diagnostic symptom category and degree of symptom improvement. Analyses of mutually exclusive client groups showed that the reduction between the baseline and the follow-up presenteeism score (WLQ-8) ranged from 27% for clients with clinical symptom severity at baseline on measures of both anxiety and depressive symptoms (n = 225) to 15% for clients who showed subclinical symptoms at baseline on measures of both anxiety and depression (n = 300). For clients with clinical levels of anxiety (n = 129) or depressive symptoms (n = 50) at baseline, significant reduction of presenteeism occurred when both reliable improvement and recovery was achieved on a respective measure of clinical symptoms. Findings suggest WMHBs can mitigate presenteeism by alleviating mental health symptoms but clients with certain symptom profiles may require greater symptom improvement to yield significant reduction of presenteeism.

Keywords:

• Workforce mental health
• presenteeism
• clinical outcomes
• employee assistance program

Introduction

Mental health problems are among the most prevalent and expensive causes of job productivity loss, specifically due to absenteeism (failure to show up as scheduled for work) and presenteeism (attending work while ill; Allen, Hines, Pazdernik, Konecny, & Breitenbach, 2018; Johns, 2010). In the United States, 30–50% of adults are expected to develop a mental illness during their lifetime (Kessler, Chiu, Demler, Merikangas, & Walters, 2005; Moffitt et al., 2010) and the economic toll of productivity loss due to depression has been estimated to be as much as $83 billion annually (Greenberg, Fournier, Sisitsky, Pike, & Kessler, 2015). Anxiety and mood disorders, including depression, are the most common types of mental illness and are associated with several forms of job productivity loss including greater absenteeism and disability days (Beck et al., 2011; Hoffman, Dukes, & Wittchen, 2008), as well as increased presenteeism and diminished functioning at work due to a health condition (Bailey, Haggarty, & Kelly, 2015).

Presenteeism may be the most expensive source of productivity loss, with some estimates indicating that costs associated with presenteeism are five to ten times greater than the costs incurred by absenteeism (Evans-Lacko & Knapp, 2016). Previous research suggests that symptoms of mental illness, particularly anxiety and mood disorders, are among the most prevalent drivers of presenteeism, (Allen et al., 2018; Garrow, 2016), as employees with a mental health problem may be more likely than those with a physical health problem to attempt to continue working despite ill health (Sanderson, Tilse, Nicholson, Oldenburg, & Graves, 2007). Recent research suggests that depressive symptoms also indirectly affect work productivity loss through subjective cognitive impairment (Toyoshima et al., 2020), which indicates the possible existence of mediating effects from other mental health impairments. These effects may be worsened by job stress specifically related to presenteeism, as a bidirectional relationship has been suggested between presenteeism and mental health distress, further compounding the negative toll on both quality of life and productivity (Demerouti, Le Blanc, Bakker, Schaufeli, & Hox, 2009).

Employers often deploy employee assistance programs (EAPs) to alleviate mental health distress and, in turn, improve job functioning and productivity across the workforce (Attridge, 2009). Available to more than half of all US employees (Bureau of Labor Statistics, 2016), the EAP typically provides brief psychosocial assessment, counseling, and care referral within an average of three sessions, at no cost to the employee (Attridge, Cahill, Granberry, & Herlihy, 2013). Although there is some overlap in interventions provided (Sharar, 2008), most EAPs differ from traditional outpatient psychotherapy services by limiting the number of available sessions to five or fewer and referring clients with problems requiring continued care to their health benefits plan or other resources (Roche et al., 2018). EAPs also differ in that their scope of services also include support for sub-clinical mental health and relationship problems, as well as workplace supports including management consultation, crisis response, and other services that address legal, financial, and family-related needs (Masi, 2020).

Despite widespread implementation of EAPs across the private and public sector (US Department of Labor, 2017) and the EAP value proposition that counseling services enhance job productivity by reducing psychological distress, there are relatively few published studies of clinical outcomes and presenteeism among EAP clients (Csiernik, 2011; McLeod, 2010).

Clinical symptom and presenteeism outcomes in traditional EAPs have been difficult to assess, in part due to the historically limited use of any systematic outcome measurement prior to more widespread adoption of the Workplace Outcomes Suite (Attridge, Sharar, DeLapp, & Veder, 2018), and in part due to EAPs’ service model itself, which emphasizes brief assessment and referral to clinical services rather than clinical treatment within the model (Masi, 2020).

Consequently, the extant literature does not fully explain how changes in clinical symptomatology during employer-sponsored counseling services affect presenteeism. In one of the most pertinent studies to-date on this topic, Richmond, Pampel, Wood, and Nunes (2016) found that among clients who received treatment in a US-based EAP, lower rates of depression and anxiety symptomatology at follow-up mediated reductions in presenteeism. However, the degree of clinical symptom improvement required to yield reduced presenteeism was not determined. Additional analyses of how treatment response affects job functioning would further clarify how employer-sponsored counseling services fulfill their value proposition and help explain how much clinical improvement is needed in order to achieve significant improvement in presenteeism.

In response to the limitations of traditional EAP models in addressing highly prevalent mental illness, as well persistent barriers to outpatient provider access in traditional health insurance networks (Melek, Davenport, & Gray, 2019), and a need for systematic assessment of clinical outcomes, enhanced models of workforce mental health benefits (WMHBs) are emerging. These benefits models are distinct from EAPs in that they provide clinical intervention across a greater number of treatment sessions and demonstrate clinical outcomes via reliable, validated measures, in line with evidence-based treatment protocols (e.g., Schneider et al., 2020).

Our retrospective cohort study examines how changes in depression and anxiety symptoms affected presenteeism among clients who received psychotherapy via a WMHB that provides a greater number of clinical sessions than traditional EAPs, vets its providers for use of evidence-based therapies (EBTs) and utilizes measurement-based care (MBC) to evaluate clinical outcomes. We hypothesized that among clients who participated in treatment through the WMHB, those with more severe symptoms would attain greater improvement in presenteeism. We further sought to understand whether changes in presenteeism differed based on the client’s symptomatology and degree of symptom improvement.

Methods

Participants

Adult clients, age 18 or older, who started individual therapy through the Lyra Health WMHB program, between September 12, 2019 and July 12, 2020, were included in the present study. The majority of the clients (73.3%) served in information technology, pharmaceuticals, and biotech industries. The rest of the clients served in industries such as transportation, retail, education, health system, and other industries. Clients must have attended at least two therapy sessions with providers contracted through Lyra Health’s clinical partners and completed two or more clinical outcome assessments, including an initial assessment within the two weeks prior to their first session, in order to be included in the study. All clients included were referred to a contracted therapist within the Lyra Health provider network. This study was deemed exempt from human participants review by the Western Institutional Review Board (WIRB).

Therapists and treatment

The 704 clients saw 467 separate licensed therapists who work in individual or group private practices as contracted partners of Lyra Health, a national WMHB program that specializes in the use of evidence-supported treatments (ESTs; Schneider et al., 2020). Prior to being admitted into the provider network with which Lyra Health partners, therapists were vetted for their self-reported practice of measurement-based care (MBC) and ESTs and interviewed to assess their knowledge of MBC and ESTs, as well as their ability to apply MBC and EST techniques for common mental health diagnoses. The median session limit in the WMHB benefits configuration was 25 sessions across employers that provided the benefit to study participants. All therapists included in the study were compensated monetarily, as per standard community practice.

Measures

The WMHB sent clients self-report measures at the start of treatment and every four weeks subsequently via secure email to assess clinical symptoms of depression and anxiety, as well as presenteeism. Depression symptoms were assessed via the Patient Health Questionnaire-9 (Kroenke, Spitzer, & Williams, 2001; PHQ-9). Anxiety symptoms were assessed via the Generalized Anxiety Disorder-7 (GAD-7) questionnaire (Spitzer, Kroenke, Williams, & Lowe, 2006). Presenteeism was assessed via the 8-item Work Limitations Questionnaire (Lerner et al., 2003; WLQ-8) which is a short-form version of the 25-item WLQ that assesses the impact of health-related limitations on work functioning in four domains: time management, physical demands, mental-interpersonal demands, and output demands. The total index score indicates the percentage of at-work time lost to presenteeism because of functional impairment on a 100-point scale. The WLQ-8 has demonstrated acceptable reliability and validity and found to be a viable alternative to the 25-item questionnaire (Walker, Tullar, Diamond, Kohl, & Amick, 2017). Therapists received client outcomes data to supplement their evaluation of client progress and inform treatment planning.

Analyses

Clients were divided into four mutually exclusive groups based on symptom and severity level: (1) subclinical, (2) anxiety, (3) depression, and (4) both anxiety and depression, according to clinical cutoffs of GAD-7 scores (GAD-7 > 7) and PHQ-9 scores (PHQ-9 > 9) that have been found to indicate symptomatology levels that meet the threshold for screening positive for an anxiety or depressive disorder, respectively (Kroenke et al., 2001; Spitzer et al., 2006).

For each subgroup, we conducted a paired t-test to evaluate whether the decreases on the Work Limitations Questionnaire-8 (WLQ-8) between baseline and follow-up were consistent across symptoms and severity levels. Furthermore, to assess whether greater improvement of clinical symptoms was associated with greater reduction in presenteeism, a 2-step multiple regression analysis was used, with the change in anxiety (GAD-7) and depression (PHQ-9) symptom scores as predictors, controlling for baseline client characteristics.

In addition, we hypothesized that the improvement in WLQ-8 score in the clinical subgroups should be greater than the improvement in the subclinical subgroup. Hence, a predictor representing the four subgroups was then implemented in the subsequent multiple regression model to determine whether there were any significant differences between the subclinical group (reference group) and three clinical subgroups.

Finally, to evaluate the differences in presenteeism reduction for clients with different clinical presentations and treatment responses, we used bootstrapped analyses, due to the small sample size of each subgroup, to obtain bootstrapped 95% confidence intervals of the reduction in the presenteeism for the clinical anxiety (GAD-7 score >7) and clinical depression (PHQ-9 score >9) subgroups. Within each clinical subgroup, we separated clients into three groups based on their treatment response (i.e., whether they obtained both reliable improvement and recovery, either, or were nonresponsive to treatment).

Reliable improvements were calculated using the RC index suggested by Jacobson and Truax (1991), which indicates whether a change in the score is greater than the measurement error of the questionnaire. We identified that reliable improvement was achieved when clients demonstrated a decrease in GAD-7 score for ≥4 points or a decrease in PHQ-9 score for ≥6 points. Clients were determined to have recovered (scores changing from clinical to subclinical range) when either their baseline GAD-7 score or PHQ-9 score were above the clinical thresholds (GAD-7 > 7, PHQ-9 > 9) and the corresponding follow-up scores were below the clinical thresholds.

For each treatment subgroup, we randomly selected samples with replacement for 1000 iterations to obtain a bootstrapped sampling distribution of the reduction in presenteeism, as indexed by the mean of differences between the baseline and the follow-up WLQ-8 scores. Next, we obtained the 95% confidence intervals from the distributions of the reduction in presenteeism for each treatment subgroup in each clinical group. If a confidence interval did not contain 0, which was the change in the WLQ-8 score claimed by the null hypothesis, we rejected the null hypothesis because it indicated that the change in the WLQ-8 score was not statistically significant. Additionally, for anxiety and depression subgroups, we performed multiple regression analyses to determine whether obtaining reliable improvement and recovery was associated with greater reduction in presenteeism. In-house scripts in R (Version 3.6.0; RStudio Version 1.2.1335) were used for all statistical analyses. Bootstrapped analyses were conducted using version 1.3-22 of the boot library in R.

Results

Of the 704 clients, 43% (n = 300) started with subclinical severity of both anxiety and depression symptoms; 18% (n = 129) started with moderate to severe anxiety symptoms; 7% (n = 50) started with moderate to severe depression symptoms, and 32% (n = 225) started in the moderate to severe range on both anxiety and depression symptoms. Groups are mutually exclusive. Table 1 reports the demographic information and the therapy attendance for the entire sample and the subgroups. On average, each therapist saw 1.5 (SD = 1.1) clients. Paired t-tests were conducted to evaluate the reduction of presenteeism between the baseline and follow-up WLQ-8 scores among clients in each respective category of symptoms and severity level (Table 2). The paired t-tests revealed that the reductions in WLQ-8 score were statistically significant across all four subgroups and the effect sizes ranged from small to moderate -0.19 to −0.55; Table 3). The results showed a 27% (-11.4/42.7, the change between the baseline and the follow-up WLQ-8 score divided by the baseline WLQ-8 score) reduction on presenteeism from clients who started with both anxiety and depression; a 26% reduction from clients who started with depressive symptoms only; a 15% reduction from clients who started with anxiety at baseline, and a 15% reduction from clients who were subclinical on both anxiety and depression at baseline.

Table 1. Demographic information and engagement with therapy.

Characteristic	Entire sample (N = 704)	Baseline subclinical^d (n = 300)	Baseline anxiety^e (n = 129)	Baseline depression^f (n = 50)	Baseline anxiety and depression^g (n = 225)
Age, mean (SD)	34.1 (9.4)	34.0 (8.8)	34.0 (9.0)	34.6 (11.4)	34.2 (9.9)
Gender, n (%)
Female	382 (54.3)	167 (55.7)	71 (55.0)	27 (54.0)	117 (52.0)
Male	322 (45.7)	133 (44.3)	58 (45.0)	23 (46.0)	108 (48.0)
Baseline GAD-7^a, mean (SD)	8.5 (5.1)	4.1 (2.0)	10.8 (2.7)	5.1 (1.8)	13.7 (3.6)
Baseline PHQ-9^b, mean (SD)	8.9 (5.4)	4.7 (2.5)	6.8 (1.9)	12.3 (2.3)	15.0 (3.9)
Baseline WLQ-8^c, mean (SD)	29.6 (20.0)	19.6 (14.7)	29.1 (17.7)	32.6 (19.3)	42.7 (19.9)
Therapy sessions completed, mean (SD)	6.5 (4.7)	6.8 (4.6)	5.4 (3.9)	6.1 (4.1)	7.0 (5.1)
Duration of care (weeks), mean (SD)	8.6 (6.7)	9.4 (6.9)	6.9 (5.4)	8.3 (6.3)	8.7 (7.1)

^aGAD-7: generalized anxiety disorder-7; ^bPHQ-9: Patient Health Questionnaire-9; ^cWLQ-8: Work Limitations Questionnaire-8; ^dGAD-7 < 8 and PHQ-9 < 10; ^eGAD-7 >=8 and PHQ-9 < 10; ^fGAD-7 < 8 and PHQ-9 >= 10; ^gGAD-7 >=8 and PHQ-9 >=10

Table 2. Change in WLQ-8 Based on symptoms and severity level.

Symptom	n	Baseline score, mean (SD)	Follow-up score, mean (SD)	Paired difference, mean (SD)	95% CI of the difference	t-Value (df)	p-Value	Cohen's d
Entire sample	704	29.6 (20.0)	23.4 (18.9)	−6.3 (17.8)	[−7.57, −4.94]	−9.33 (703)	<.001	−0.32
Subclinical	300	19.6 (14.7)	16.8 (14.9)	−2.8 (14.2)	[−4.46, −1.23]	−3.47 (299)	<.001	−0.19
Anxiety	129	29.1 (17.7)	24.8 (16.6)	−4.3 (16.3)	[−7.13, −1.45]	−2.99 (128)	.003	−0.25
Depression	50	32.6 (19.3)	24.1 (19.2)	−8.6 (20.8)	[−14.47, −2.66]	−2.92 (49)	.005	−0.45
Anxiety and depression	225	42.7 (19.9)	31.3 (21.5)	−11.4 (20.8)	[−14.15, −8.68]	−8.23 (224)	<.001	−0.55

Table 3. Multiple regression analysis of the change of WLQ-8 presenteeism (n = 704).

	Model 1		Model 2
Predictors	Estimates (95% CI)	p-Value	Estimates (95% CI)	p-Value
Baseline WLQ-8	−0.36 (−0.41 – −0.31)	< .001	−0.47 (−0.53 to −0.42)	<.001
Session count	0.05 (−0.18 − 0.27)	.677	0.11 (−0.10 − 0.32)	.309
Age at initial session	0.04 (−0.07 − 0.15)	.476	0.04 (−0.06 − 0.14)	.447
Gender [male]	1.19 (−0.89 − 3.26)	.261	1.47 (−0.49 − 3.43)	.140
Change in GAD-7	0.86 (0.57 − 1.14)	< .001	1.17 (0.87 − 1.48)	<.001
Change in PHQ-9	0.73 (0.45 − 1.02)	< .001	1.01 (0.70 − 1.32)	<.001
Baseline symptom [anxiety]			7.55 (4.58 − 10.53)	<.001
Baseline symptom [depression]			5.16 (0.89 − 9.43)	.018
Baseline symptom [anxiety and depression]			14.04 (11.05 − 17.02)	<.001

A multiple regression analysis was performed on the full sample with the change in WLQ-8 presenteeism scores as the dependent variable and the change in GAD-7 anxiety score and the change in PHQ-9 depression score as predictors, controlling for age, gender, number of sessions, and baseline WLQ-8 score (Table 3). The results showed that there were significant positive effects between the change in WLQ-8 score and both the changes in GAD-7 and PHQ-9 scores. These findings indicated that an improvement in WLQ-8 score was associated with improvements in both anxiety score and depression score. Building on these findings, we further compared the previous model to a model with an additional predictor of clinical subgroups to test the main effect of the subgroups. The model showed a significant main effect of subgroups (χ² = 84.34, df = 3, p < .001). The results indicated that clients who started in any clinical subgroups showed significantly less reduction in presenteeism given the same amount of improvement in GAD-7 and PHQ-9 scores (Table 3).

Anxiety symptoms

Of the 704 clients, 129 (18%) scored in the clinical range on GAD-7 at baseline. Among this group, the average baseline score on GAD-7 was 10.8 (SD = 2.7), corresponding to the moderate range of severity. Table 4 shows the bootstrapped 95% confidence intervals for the mean of the differences between the baseline and the follow-up WLQ-8 scores by the response status on the GAD-7 score. The results indicated that only the clients with both reliable improvement and recovery showed a statistically significant improvement on presenteeism score (WLQ-8). Furthermore, we applied a multiple regression model to predict presenteeism (i.e., the change in WLQ-8 scores) from whether clients attained both reliable improvement and recovery, controlling for baseline characteristics. The results of the multiple regression model showed that having both reliable improvement and recovery on the GAD-7 score was positively associated with a greater improvement in presenteeism, and the effect was statistically significant ($\beta$= −10.86, SE = 2.28, t = −4.76, p < .001).

Table 4. Bootstrapped confidence intervals for the change in WLQ-8 based on symptoms and response statuses.

	Anxiety				Depression
Response Status	n	Baseline score, mean (SD)	Follow-up score, mean (SD)	Bootstrapped 95% CI	n	Baseline score, mean (SD)	Follow-up score, mean (SD)	Bootstrapped 95% CI
Nonresponse	37	31.2 (17.7)	33.4 (15.8)	[−3.02, 7.24]	12	37.4 (19.2)	40.2 (21.4)	[−3.27, 9.56]
Either reliable improvement or recovery	21	30.3 (18.7)	30.1 (18.7)	[−5.72, 5.74]	10	32.9 (20.5)	23.0 (14.6)	[−22.69, 6.26]
Both reliable improvement and recovery	71	27.7 (17.5)	18.8 (13.9)	[−12.45, −5.27]	28	30.5 (19.3)	17.5 (15.8)	[−20.03, −4.91]

Depression symptoms

At baseline, 50 clients (7%) scored in the clinical range on the PHQ-9. Among this group, the average baseline score on PHQ-9 was 12.3 (SD = 2.3), corresponding to the moderate range of severity. The bootstrapped 95% confidence intervals indicated that for clients with depression symptoms at baseline, having both reliable improvement and recovery was significantly associated with an improvement in presenteeism (Table 4). In addition, we conducted a multiple regression model to predict presenteeism (i.e., the change in WLQ-8 scores) from whether clients obtained reliable improvement and recovery at the end of the treatment, controlling for baseline characteristics. Results of a multiple regression model revealed that clients who had a reliable improvement and recovered from depression showed a statistically significant improvement in presenteeism ($\beta$= −13.38, SE = 5.38, t = −2.49, p=.017).

Discussion

Employers invest in workforce mental health solutions to relieve the burden of behavioral health conditions on workforce productivity, quality of life and overall health (Attridge, 2009). Consequently, it is important to understand how these benefits solutions deliver on their value proposition under naturalistic conditions. While prior studies suggest that EAPs can significantly improve mental health-related outcomes and productivity (McLeod, 2010), research remains sparse regarding differences in treatment response based on clinical symptom type, and how treatment responses impact costly productivity loss. Further, research examining outcomes from emerging models of workforce mental health benefits (WMHB) programs that offer a greater number of sessions in service of providing robust clinical treatment, is still nascent.

Our results support the limited prior findings from EAPs, showing that baseline depression and anxiety symptoms were positively associated with increased presenteeism (Aronsson & Gustafsson, 2005; Collins et al., 2005), while also corroborating the extant research that demonstrates decreased clinical symptoms and presenteeism during counseling episodes (Richmond et al., 2016). Additionally, our study expands upon earlier research by demonstrating how symptom improvement predicts change in presenteeism.

Presenteeism decreased for clients who saw improved depression or anxiety symptoms, regardless of baseline symptom severity or comorbidity, and this effect was predicted by the degree of clinical improvement during the course of treatment delivered via the WMHB. In addition, for clients who started with more severe symptoms at baseline, they required a greater degree of improvement in their clinical symptoms to attain the same level of improvement in presenteeism. The results of the multiple regression on the full sample, controlling for the baseline characteristics, showed that a unit of improvement in the depressive score (PHQ-9) is 16% less effective in reducing presenteeism than a unit improvement in the anxiety (GAD-7) score (Table 3). These findings are congruent with prior research showing that among mental illnesses, depression is especially detrimental to work productivity due to a constellation of symptoms that often impair performance, such as loss of motivation, lethargy, and diminished concentration. Thus, relative to anxiety symptoms, greater reduction in depression symptoms may be needed to achieve a similar reduction in presenteeism. Although anxiety has also been shown to negatively affect productivity (Hoffman et al., 2008), studies indicate that the effects may be nuanced and that mild anxiety can actually facilitate improved work performance and productivity, depending on diverse factors including employee, job, and situational characteristics (Cheng & McCarthy, 2018). Overall, our findings suggest that although the negative impact of anxiety and depression symptoms on presenteeism can remit during treatment via the WMHB, reductions in presenteeism are partly based on which types of clinical symptoms decrease and to what degree.

Given the value proposition of employer-sponsored mental health benefits, these findings may have implications for the design and evaluation of such solutions. For example, understanding differential treatment response on presenteeism, one of the most expensive forms of productivity loss, could influence the configuration of WMHB models to ensure that offerings are robust enough to yield greater treatment response. These results also demonstrate both the economic and clinical value of systematic outcomes data collection, both for the employer to understand their return on investment, and for the benefits program to understand the effectiveness of their provider network.

Several limitations should be noted. The strong external validity of our naturalistic cohort study design also means that our ability to make causal inferences is diminished, as subjects were not randomly assigned to different conditions. Thus, the current study does not determine causal relationships between the change in clinical symptoms and the change in productivity. Secondly, because we could not conduct an intent-to-treat analysis, it’s unclear how our sample may differ from clients who prematurely ended treatment or declined to complete outcome measures. Aside from basic client demographics, we were unable to assess the impact of specific client, provider, or treatment characteristics due to the limitations of our dataset. Furthermore, because clients completed follow-up outcome measures at variable time points, and follow-up data may reflect outcomes prior to end of treatment, it is likely that actual pre-post treatment effects are underestimated by reported results (Hansen, Lambert, & Forman, 2002; Schneider et al., 2020).

Although our findings demonstrate that depression, anxiety, and presenteeism significantly improved during care episodes, our findings may not generalize to all WMHB programs due to the unique characteristics of this particular program, including its emphasis on the delivery of EBTs, use of MBC, and offering of more treatment sessions than in a standard EAP model of six or fewer sessions (Schneider et al., 2020).

Our study indicates several potential areas for future research to examine the effectiveness of WMHB programs in improving presenteeism and mental health outcomes. While our study only analyzed baseline and final outcome measures, subsequent studies should utilize a longitudinal approach to understand the full extent and long-term durability of treatment effects. Future research should also employ quasi-experimental designs to evaluate the effects of WMHB programs on presenteeism, absenteeism, and mental health symptoms, as well as the mediating and moderating relationships between these variables. Study designs using randomization and control groups would enhance methodological rigor and enable causal inferences about program effectiveness, which has historically been lacking in even the extant EAP research literature (Richmond, Pampel, Wood, & Nunes, 2017). Further, researchers should evaluate how presenteeism improvements during care episodes may vary within the subclinical population based on mental health outcomes other than depression and anxiety. For example, improvements on non-diagnostic measures of mental health (e.g., perceived life stress, wellbeing) during the course of treatment may better account for changes in presenteeism among subclinical clients. Finally, additional research is needed to understand how particular treatment protocols, client and provider characteristics, and care model configurations may differentially affect clinical and productivity outcomes, as such findings could enhance the design of care services to optimize client outcomes.

Disclosure statement

All authors were employed by Lyra Health at time of authorship and received commensurate compensation for their work as full-time, salaried employees. No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.