Depression treatment initiation among patients with versus without chronic pain

Abstract

Objective:

The purpose of this study was to examine the extent to which the presence of chronic non-cancer pain (CNCP) impacts the likelihood that patients with diagnoses of depression will initiate depression treatment compared to those without CNCP.

Methods:

We performed a retrospective cohort study of Kaiser Permanente of Georgia members ≥18 years who received a diagnosis of depression. Demographics and medical history were extracted from the electronic health record database. Members were further classified by the presence or absence of a CNCP diagnosis. Outcomes of interest were treated as time dependent and included (1) time to fulfillment of a new anti-depressant medication and (2) time to a follow-up mental health encounter. Outcomes were compared between members with and without a CNCP diagnosis using Kaplan-Meier survival curves and Cox Proportional hazard regression models.

Results:

During the study period, 22,996 members met inclusion and 27.4% had a diagnosis of CNCP. In the matched sample, there was no difference in the time to a new anti-depressant fill among members with and without CNCP (HR=0.96; 95%CI: 0.90, 1.02; p=0.18). In contrast, members with CNCP were significantly less likely to have a new mental health encounter following diagnosis (HR: 0.87; 95%CI: 0.81, 0.94; p<0.001).

Conclusions:

Patients with CNCP were significantly less likely to have a new mental health encounter following a depression diagnosis compared to patients without CNCP. Additional outreach and consideration may be needed to improve initiation of depression treatment for newly diagnosed patients with comorbid depression and CNCP.

INTRODUCTION

Depression and chronic, non-cancer pain (CNCP) comorbidity is incredibly common, with depression symptomatology as high as 50% among patients with certain CNCP conditions (1) and the prevalence of pain symptoms in populations with depression as high as 77% (2–4); thus, this comorbidity is often termed as a “depression-pain syndrome” or “depression-pain dyad” because of the frequency with which these conditions co-occur, the ways in which they can exacerbate each other, and their shared neurobiology including common brain structures (e.g., prefrontal cortex, thalamus) and neurotransmitters (e.g., serotonin, norepinephrine, glutamate) (2, 5).

Unfortunately, evidence consistently suggests that this comorbidity has serious implications with respect to both condition severity and treatment efficacy. For example, individuals with depression are more likely to report more intense/more severe pain symptoms (6, 7) and are more likely to experience a longer duration of persistent pain (8–11), non-recovery from painful conditions (9) and relapse following treatment compared to those without depression (12, 13). Similarly, individuals with CNCP pain are more likely to report more severe depression (14), have a worse response to antidepressants (15–17) and have a longer time to depression remission compared to individuals without pain (18–22).

What is still unknown, however, is whether the presence of CNCP impacts the likelihood that patients with diagnoses of depression will even initiate depression treatment. In their meta-analysis estimating the likelihood of spontaneous remission from untreated depression, Whiteford et al. reported that close to half of prevalent cases of untreated major depression will not remit spontaneously in a given year; these rates are even lower (by ~20–30%) among those with more severe depression (23). Given the increased risk of more severe depression symptomatology among individuals with CNCP and the human cost to untreated depression, including protracted suffering, impairments in work and interpersonal relationships, and heightened risk of suicide (24), it is critical to assess treatment initiation patterns in this population. Thus, the purpose of this study was to examine the extent to which the presence of CNCP impacts the likelihood that patients with diagnoses of depression will initiate depression treatment compared to those without CNCP.

Methods

Setting

Kaiser Permanente of Georgia (KPGA) is an integrated healthcare delivery system offering both health insurance coverage and health care services at 25 outpatient facilities throughout the metropolitan Atlanta area. Individuals can enroll directly in private individual/family plans, as part of Medicaid (for eligible children only) or can enroll through employer-based plans. Medicare Advantage options are available through both individual and employer-based plans for those eligible (age ≥65, permanently disabled and receiving social security benefits, diagnosed with end-stage renal disease and/or have ALS). The KPGA membership is racially diverse (~44% Black/African-American, ~32% White, ~6% Asian), 54% female, has a median age of 41 (range=0–104, SD=20.71) and a median neighborhood household income of $69,966 (range=$7,861-$250,001, SD=$29,243.53).

Study Population and Outcomes of Interest

We performed a retrospective cohort study of KPGA members ≥18 years who received a new diagnosis of depression [International Classification of Disease (ICD)-10 F06.31, F06.32, F32.0, F32.1, F32.2, F32.3, F32.4, F32.5, F32.8, F32.81, F32.89, F32.9, F33.0, F33.1, F33.2, F33.3, F33.40, F33.41, F33.42, F33.8, F33.9, F34.1, F43.21, F43.23, F53., F53.0, F53.1) between 10/1/2016–10/1/2019 following a 365-day period without evidence of a depression diagnosis or treatment [psychotherapy or anti-depressants (AD)]. Eligible individuals had to have continuous health plan membership (but could have a gap in enrollment records of ≤30 days, as administrative gaps can occur as a result of delays in membership data processing and thus are not indicative of membership interruptions/disenrollment); patients were excluded if disenrolled from the healthcare system within 90 days after their depression diagnosis.

Eligible patients were classified by the presence or absence of a CNCP diagnosis (≥ 2 CNCP diagnoses in the year prior to the depression diagnosis). The chronic pain conditions extracted included: back pain, neck pain, limb/extremity pain, arthritis, fibromyalgia/widespread muscle pain, headache, orofacial/ear/temporomandibular pain, abdominal/bowel pain, chest pain, urogenital/pelvic/menstrual pain, fractures/contusions/sprains/strains and other painful conditions [which included sickle cell disease, complex regional pain syndrome, systemic lupus erythematosus, acquired deformities (excluding spinal disorders), spinal cord injury and neuropathic pain]. Pain documented as cancer-related pain was excluded. The list of ICD codes used for identifying pain conditions are available online (https://github.com/MHResearchNetwork/MHRN-Central/blob/master/WP_MHRN_SMI_painOpioids.zip).

Time to and occurrence of depression treatment initiation was defined as (1) a filled antidepressant (AD) medication prescription or (2) at least one completed psychotherapy visit (a visit > 30 minutes in duration to a specialty mental health provider with a Current Procedural Terminology (CPT^®) code indicating ‘initial evaluation’ or ‘individual psychotherapy’) within 12 months after the CNCP diagnosis date. See the Supplemental Digital Content for a list of AD medications included.

We also examined sociodemographic (age, sex, race/ethnicity, neighborhood socioeconomic status) and clinical characteristics of the study population based on data from the year prior to the depression diagnosis using methods similar to prior work (25). Overall medical comorbidity burden was calculated using the Charlson Comorbidity Index Score (CCIS). This score consists of 19 categories of comorbidity, with each category weighted based on the adjusted risk of 1-year post-discharge mortality. The overall comorbidity score reflects the cumulative increased likelihood of mortality one year after discharge such that higher scores are indicative of a more severe burden of comorbidity (26). Total health care utilization (hospitalizations, ED visits and other in-person outpatient encounters) was based on summarized encounter data from the year prior to the depression diagnosis. This timeframe was selected so that we had a baseline measure of recent utilization history prior to the study period. Multiple encounters occurring on the same day were coded as a single encounter so that we were able to count utilization days. The Institutional Review Board at Kaiser Permanente Georgia approved the study protocol for this project with a waiver of informed consent.

Statistical Analyses

Descriptive statistics were calculated for all variables of interest and included means and standard deviations or medians and ranges for continuous variables or counts and percentages for categorical variables. Demographics and medical history data were compared between patients with and without CNCP diagnosis using Chi-square tests for categorical data and two-sample t-tests or Wilcoxon rank sum tests for continuous data. Given that demographic and medical history data may differ between patients with and without a CNCP diagnosis, we chose to a priori control for possible confounders using a propensity score matching approach. Variables included in the propensity score were gender, race/ethnicity, age at depression diagnosis, baseline Patient Health Questionnaire (PHQ-9) score as a measure of depression severity, Charlson Comorbidity score, median neighborhood household income and number of healthcare encounters in year prior to depression diagnosis. The propensity score was constructed using multivariable logistic regression models using the variables described above. Matching patients with and without a CNCP diagnosis was performed at a 1:1 ratio using a greedy nearest neighbor matching algorithm and implemented using the PSMATCH procedure in SAS(27). To assess the quality of matching, the distribution of variables included in the propensity score were examined after matching.

Outcomes of interest were treated as time dependent and included (1) time to fulfillment of a new AD medication and (2) time to a new mental health encounter. Only patients with a new order of an AD medication were considered for the medication fulfillment outcome. For both outcomes, patients were censored after 12 months or health plan disenrollment (>90 days from depression diagnosis). In addition, we examined time to any new depression treatment initiation which captured either of these events. Outcomes were compared between members with and without a CNCP diagnosis using Kaplan-Meier survival curves and Cox Proportional hazard regression models. The proportional hazard assumption was assessed in the propensity matched sample using visual inspection and tests of the Schoenfeld residuals. To control for the propensity matched pairs in model estimation, stratified Cox-PH models were constructed and the overall “common” hazard ratio is presented with associated 95% confidence intervals. Analyses were conducted using SAS Enterprise Guide version 8.2 (SAS Institute, Cary, NC) and statistical significance was assessed at the 0.05 level, unless otherwise noted.

Results

During the study period, 22,996 members met inclusion and 27.4% had a diagnosis of CNCP. Members with a depression diagnosis were predominately female (69%), Black/African American (43%) or White (39%), with an average age of 42.9 (SD=16.1) years at time of diagnosis (Table 1). Prior to matching, patients with a CNCP diagnosis tended to be older (mean age 49.6 vs. 40.4 years), Black/African American (53% vs. 40%), have higher comorbidity scores (mean 1.1 vs. 0.5), and higher healthcare utilization in prior year (8 visits vs. 1 visit). Among the 22,996 patients, 8,758 (38.1%) had a new MH encounter during the follow-up period. In addition, 15,856 (69.0%) had an anti-depressant order, of which 13,904 (87.7%) of patients filled their AD order. Patients with a diagnosis of CNCP were propensity matched to patients without chronic pain at a 1:1 ratio. Among the 6,307 patients with CNCP, 5,368 (85%) were matched to patients without chronic pain. Compared to matched patients, unmatched patients were more likely to be older, male, have more comorbidities, have a lower average income, and more than two times the number of encounters in prior 12 months. See Table S1, Supplemental Digital Content, for additional details.

Table 1:

Summary of Overall Cohort stratified by those with and without chronic pain (Complete Case Sample n = 22,996)

	Overall	Depression with Chronic Pain N = 6,307	Depression without Chronic Pain N = 16,689	P-value
Age at Depression Diagnosis Mean +/− SD [Median]	42.9 (16.1) [42]	49.4 (16.1) [50]	40.4 [38]	<.001
Race				<.001
Asian	742 (3.2%)	200 (3.2%)	542 (3.3%)
Non-Hispanic Black/African American	9,978 (43.4%)	3,327 (52.8%)	6,651 (39.9%)
Other	242 (1.1%)	68 (1.1%)	174 (1.0%)
Non-Hispanic White	8,957 (39.0%)	2,096 (33.2%)	6,861 (41.1%)
Unknown	3,077 (13.4%)	616 (9.8%)	2,461 (14.8%)
Gender- Female	15,911 (69.2%)	4,413 (70.0%)	11,498 (68.9%)	0.116
Median Household Income Mean +/− SD [Median]	69,586 (27,650) [66,625]	66,486 (26,034) [64,176]	70,757 (28,149) [67,279]	<.001
Charlson Score	0.49 (1.3) [0]	1.06 (1.82) [0]	0.28 (0.95) [0]	<.001
Baseline PHQ Score (n = 11,319) Mean +/− SD [Median]	14.2 (6.0) [14]	14.3 (5.8) [14]	14.2 (6.04) [14]	.140
< 10	2,595 (11.3%)	624 (9.9%)	1,971 (9.9%)
≥ 10	8,724 (37.9%)	2,282 (36.2%)	6,442 (38.6%)
Unknown	11,677 (50.8%)	3,401 (53.9%)	8.276 (49.6%)
Number of Healthcare Encounters in Prior 12 Months Mean +/− SD [Median]	5.1 (11.0) [3]	11.0 (15.7) [8]	2.8 (7.4) [1]	<.001

In the matched sample, 6,881 members (64.1%) had a new order for an AD medication. Among them, 5,850 (85.0%) filled the medication order. When examining medication fill as a time-dependent outcome (Figure 1), there was no difference in the time to a new AD fill among members with and without CNCP (HR=0.96; 95%CI: 0.90, 1.02; p=0.18). In contrast, members with CNCP were significantly less likely to have a new mental health encounter following diagnosis (HR: 0.87; 95%CI: 0.81, 0.94; p<0.001; see Table 2). At 4 months, 30.1% of patients with CNCP had a new mental health encounter compared to 33.8% of patients without a CNCP diagnosis (Figure 2). At 7 months, 33.4% of patients with CNCP had a new mental health encounter compared to 36.6% of patients without a CNCP diagnosis (Figure 2). When combining events (encounter or prescription fill) to evaluate treatment initiation, there was no difference in time to treatment initiation among members with and without CNCP (HR: 1.00; 95%CI: 0.95, 1.06; p=0.187).

Figure 1.

KM curve med fill (matched sample but not accounting for matched pairs AND has a MED Order (reduced sample))

Table 2:

Adjusted Hazard ratios for each of the 3 outcomes (matched sample accounting for matched pairs)^{^}

Outcome	Hazard Ratio*	95% CI	P-value
New Depression Medication Fill (conditional on med order)	0.98	(0.93, 1.05)	0.665
New Mental Health Encounter	0.87	(0.81, 0.94)	<0.001
New Medication Fill or Mental Health Encounter	1.00	(0.95, 1.06)	0.989

^{^}Variables included in the propensity score were gender, race/ethnicity, age at depression diagnosis, baseline Patient Health Questionnaire (PHQ-9) score as a measure of depression severity, Charlson Comorbidity score, median neighborhood household income and number of healthcare encounters in year prior to depression diagnosis.

^*Patients without chronic pain are in the reference group

Figure 2.

KM curve new MH visit (matched sample but not accounting for matched pairs)

Discussion

The results of this study suggest that patients with CNCP were significantly less likely to have a new mental health encounter following a depression diagnosis compared to those without CNCP. Riva et al. argue that pain requires attention, which consequently depletes cognitive resources and evidence does suggest that individuals with chronic pain often report cognitive impairments such as diminished concentration, poor memory performance, and slower cognitive processing (28, 29). Thus, patients with CNCP may not have the bandwidth for scheduling and keeping a mental health-related appointment, which takes initiative and additional time/effort. This, in combination with depressive symptoms – which often include social withdrawal (30) – may partly explain the lower likelihood of mental health encounters among individuals with CNCP following a depression diagnosis.

Further, because of the high attentional demand of pain, it can be hard for individuals to focus on anything but the presence of pain (31). Indeed, in one study 60% of CNCP patients reported they had visited their doctor between 2–9 times in the months prior with 11% having done so at least 10 times (32). Therefore, while patients may be over-utilizing healthcare resources for issues related to their pain, they may be under-utilizing other services including specialty mental healthcare. Patients with comorbid depression and pain and their providers may also be more focused on pain- as opposed to depression-related issues during clinical encounters. In their expansion of the Competing Demands Model, Nutting and colleagues highlight that patient perceptions and preferences impact the likelihood of depression treatment for individuals with physical comorbidities. For example, if a patient perceives his/her depression to be less severe or a result of a ‘legitimate medical concern’ (e.g., pain), he/she may be less likely to initiate depression treatment (33). Providers may face challenges with how to effectively address multiple concerns during a single, time-constrained encounter and thus prioritize treating pain-related complaints; they may also be less likely to emphasize with patients the importance of scheduling a psychotherapy visit if they think that depression is an “understandable” reaction to physical illness (e.g., pain) (34). Due to the fact that patients do not need a formal referral to schedule follow-up specialty mental healthcare appointments in the KPGA system, we were not able to assess whether providers were less likely to refer patients with versus without CNCP for mental healthcare and, though there have been some recent studies focused on better understanding factors that influence provider mental healthcare referrals (35), this is an important domain for future research particularly among patient populations with comorbid pain and depression.

Patients with CNCP were not less likely than those without CNCP to fill a new AD prescription. This may be partly explained by the ease with which patients can fill a prescription in the KPGA system, where all 25 medical office facilities in Metro Atlanta have in-house pharmacies and prescriptions can be available while patients wait. Also, given that some AD medications can also be prescribed for pain management (36), patients with CNCP in our study may have been more motivated to fill the prescription (assuming providers communicated the dual benefit of these medications during the encounter). However, findings from a recent Australian cohort study suggest that participants with comorbid chronic pain and depression reported fewer functional benefits from AD use and lower benefits from sertraline, escitalopram, and venlafaxine compared to participants without chronic pain (37), underscoring the importance of psychotherapy visits instead of or as a complement to AD for this population.

The present study has several limitations. First, diagnostic codes for pain do not provide information about pain severity. Second, administrative data do not provide information about the reasons for the antidepressant prescription; tricyclic antidepressants are commonly prescribed for pain and/ or depression (36). Third, mental health encounters could have been initiated for reasons other than depression. Finally, study results were derived from a sample of members of an integrated payer–provider system. Thus, caution is urged in generalizing the findings to uninsured populations. For example, because evidence consistently suggests that individuals with depression are more likely to lose their health insurance and less likely to obtain health insurance if uninsured(38–40), our insured study population may not be representative of all individuals with depression and perhaps could have less severe depression than those not able to initiate/maintain health insurance coverage. Uninsured individuals with depression could also be even less likely to initiate depression treatment compared to insured individuals due to the additional cost associated with psychotherapy visits or anti-depressant medications. Strengths of the study include a large, geographically and racially/ethnically diverse study population, a matched sample, and the controlling of possible confounders using propensity scores.

Conclusions

In spite of a great deal of progress in pain treatments over recent decades, these treatments help some, but not all, CNCP patients; this may be due to the fact that there are other comorbidities including depression that are interfering with the effectiveness of pain treatment. A growing body of research supports this hypothesis: one recent randomized controlled trial found that an early reduction in depressive symptoms was associated with a reduction in pain interference but an early reduction in pain interference was not associated with reductions in depressive symptoms (41). Similarly, another study reported that initial depression treatment subsequently reduced pain interference among individuals with comorbid chronic pain and depression (42). Therefore, prioritizing treatment of depression in this population is critical for both improving depression- and pain-related outcomes. Additional outreach and consideration may be needed to improve initiation of depression treatment for newly diagnosed patients with comorbid depression and CNCP.

List of Abbreviations:

CNCP

Depression and chronic, non-cancer pain

KPGA

Kaiser Permanente of Georgia

AD

Antidepressant

ICD

International Classification of Disease

CPT^®

Current Procedural Terminology

CCIS

Charlson Comorbidity Index Score

PHQ-9

Patient Health Questionnaire