Under-documentation of Psychiatric Diagnoses among Persons Living with HIV in Electronic Medical Records

Abstract

While psychiatric disorder prevalence is higher among patients living with HIV (PLWH), there are no prior studies on rates of psychiatric diagnosis documentation in electronic medical records (EMRs) of PLWH, which was the goal of this study. Participants (n = 2,336) were enrolled in the Center for AIDS Research (CFAR) Longitudinal Database study. Diagnostic codes were extracted to assess the prevalence rates of depressive, alcohol/substance use, PTSD, sleep, and adjustment disorders. Prevalence rates for psychiatric disorders were extracted from national epidemiological studies for comparison. Rates of Major Depressive Disorder in the EMR were comparable to prior epidemiological reports in PLWH. In contrast, rates of PTSD, substance use, alcohol use, adjustment, and insomnia disorders were all markedly lower compared to epidemiological studies. While clinicians appropriately documented depression, other psychiatric comorbidities were largely overlooked. This indicates a potential bias in the detection or documentation of psychiatric disorders in PLWH.

INTRODUCTION

Persons living with HIV (PLWH) are significantly more likely to meet criteria for a psychiatric diagnosis compared to the general population (1–2). While prevalence estimates vary, up to 50% PLWH are estimated to meet criteria for a psychiatric disorder in their lifetime (3). The most common psychiatric comorbidities among PLWH are substance use disorders, depression, posttraumatic stress disorder (PTSD), and anxiety disorders (4).

The consequences of psychiatric comorbidity among PLWH are serious. Psychiatric comorbidities are associated with poor adherence to anti-retroviral therapies (ART), which reduces the likelihood of achieving viral load suppression (5–6). PLWH who have a psychiatric comorbidity are more likely to die by suicide compared to PLWH without a psychiatric diagnosis (7). The presence of a psychiatric diagnosis contributes to early all-cause mortality among PLWH (8). Finally, psychiatric comorbidities among PLWH may contribute to the enormous economic burden association with HIV infection (9).

Fortunately, evidence-based practices (EBPs) are extremely effective at reducing psychiatric symptoms among PLHW, thereby reducing the burden of illness. For example, in a randomized controlled trial, Safren and colleagues (10) found that cognitive behavioral therapy (CBT) led to significantly greater reductions in depression and significantly better treatment adherence relative to treatment as usual in PLWH. Similarly, PLWH had significantly greater reductions in PTSD symptoms in response to prolonged exposure (PE) therapy for PTSD compared to a monitoring condition and these gains were maintained up to 6 months after treatment termination (11). In terms of substance use symptoms, a pilot of PLWH who had a substance use disorder found that CBT was associated with significant reductions in alcohol use and depression (12). Therefore, CBT is effective for a variety of psychiatric comorbidities among PLWH.

While there is overwhelming evidence that psychiatric conditions can be mitigated among PLWH, receipt of evidence-based practice is contingent upon accurate diagnosis. In busy and under-resourced medical settings, a variety of logistical and psychological factors may prevent the detection and documentation of psychiatric disorders. For instance, the stigma of mental illness (13) may dissuade a provider from documenting a psychiatric disorder among PLWH, who are already at higher risk for stigma (14). Without a psychiatric diagnosis, PLWH are unlikely to receive a referral for EBPs or to recover from their illness (15). It is essential to understand the frequency of detection and documentation of mental health conditions among PLWH in clinical practice.

Electronic medical records (EMRs) provide a useful tool for understanding documentation of psychiatric illness among PLWH in care. While national epidemiological studies demonstrate that psychiatric disorder prevalence is higher among PLWH, to our knowledge there are no prior studies that report rates of psychiatric diagnoses documentation in EMRs of PLWH. This is a critical gap in the literature that requires research to inform implementation of evidence-based practice for psychiatric disorders of PLWH.

The goal for this study was to report the rates of a variety psychiatric diagnoses among PLWH in EMRs. More specifically, we sought to explore the frequency of diagnosis of major depressive disorder, posttraumatic stress disorder (PTSD), insomnia, alcohol use disorders, and substance use disorders in EMRs for a large sample of PLWH. These disorders were selected because they are some of the most commonly reported psychiatric conditions among PLWH (16–18). We then aimed to compare these rates to prevalence estimates of these psychiatric conditions from epidemiological studies of the general population and of PLWH. Based on prior research on stigma of psychiatric diagnoses (13), we hypothesized that rates of documentation of psychiatric disorders would be lower in the EMR compared to epidemiological estimates of PLWH and the general population.

METHODS

Participants.

Participants (n = 2,336) were participants in the Center for AIDS Research (CFAR) Longitudinal Database study at the University of Pennsylvania, Philadelphia, PA. Demographic information and diagnostic code frequencies are reported in Table I. All CFAR Participants in this study were HIV positive and provided informed consent allowing for the extraction of their EMR data.

Table I.

Demographic Information

	n (%)
Gender
Male	1,704 (73.0)
Female	632 (27.1)
Age, M (SD)	52.8 (11.4)
Marital Status
Single	1,798 (77.0)
Separated	56 (2.4)
Divorced	121 (5.1)
Widowed	48 (2.1)
Married	313 (13.4)
Ethnicity
Black/African American	1,474 (63.1)
White	776 (33.2)
Asian	9 (0.4)
Native Hawaiian/Pacific Islander	2 (0.0)
Other	48 (2.1)
“Unknown”	26 (1.1)
Missing	1 (0.0)
Race
Hispanic/Latino	104 (4.5)
Non-hispanic/Non-Latino	2,167 (92.8)
Missing	65 (2.8)
Psychiatric Disorders
Adjustment-Related Disorders (309.9, F43.20–29)	103
Stress-Related Disorders/PTSD (309.81, F43.0–10)	98
Alcohol-Related Disorders (F10.10–951)	168
Cannabis-Related Disorders (F12.10–90)	34
Cocaine-Related Disorders (F14.10–929)	136
Hallucinogen-Related Disorders (F16.10–929)	7
Depression (301.12, F32.0–34.1)	771
Opioid-Related Disorders (F11.10–229)	81
Other Psychoactive Substance Related Disorders (F19.10–981)	91
Sedative-Related Disorders (F13.10–239)	10
Sleep-Related Disorders (F51.01–12, G47.00–419, Z72.280)	499

Note: Psychiatric disorder frequency include participants who met criteria for multiple diagnoses within a category (i.e., multiple sleep-related disorders) as well as those who only met one within a category. Therefore these values cannot be used for prevalence rates. However, Table II data reports frequencies with any dual-diagnosis individuals counted only once per category as an indication of prevalence.

Diagnostic Codes.

Diagnostic codes were extracted to assess the prevalence of depressive disorders, alcohol and substance use disorders, PTSD, sleep disorders, and adjustment disorders (see Supplemental Table I) for all CFAR participants.

Prevalence Rate Comparisons.

Prevalence rates for psychiatric disorders were extracted from national epidemiological studies for these diagnoses. All studies used for prevalence rate comparisons are cited in the footnote for Table II.

Table II.

Prevalence Rates in National Surveys in the General Population, HIV positive population, and in the electronic medical record.

Disorder	General Population Lifetime Prevalence (%)	HIV+ Population	Electronic Medical Record Prevalence
Major Depression	16.2	15.8–22	19.3 (Including other depressive disorders: 20.6)
PTSD	6.8	30–42	2.6 (Including other trauma disorders: 3.5)
Adjustment Disorders	1–2	25.9	3.9
Insomnia	6.0	38-47-67-73	7.5 (Including other sleep disorders: 14.1)
Alcohol Abuse	13.2	8–10.9	2.4 (Including all alcohol use disorders: 4.5)
Substance Abuse	7.9	19.6–21	3.3 (Including all substance use disorders: 7.6)

Note. Prevalence rates from (2, 31–46)

Extraction Procedures.

The University of Pennsylvania maintains a research repository of electronic health information that is compiled from multiple University of Pennsylvania Health System (UPHS) clinical information databases and is updated daily. This repository was used to query all hospital encounter records for participants enrolled in the CFAR registry to identify occurrences of the diagnoses of interest. Extracted data sets identified participants using only CFAR registry participant IDs to protect participant confidentiality.

RESULTS

Compared to the general population, rates of Major Depressive Disorder and insomnia were slightly higher in the sample from the EMRs. In contrast, PTSD, substance use disorders, alcohol use disorders, and adjustment disorders were all lower in the EMR sample compared to national prevalence estimates for the general population.

Rates of Major Depressive Disorder (MDD) in the electronic medical record were comparable to prior reports on HIV-infected samples. In contrast, rates of PTSD, substance use disorders, alcohol use disorders, adjustment disorders and insomnia from the EMR were all reported at markedly lower rates compared to prevalence estimates for HIV-infected samples in the literature.

DISCUSSION

Documentation of most psychiatric disorders was lower in EMRs of PLWH compared to population and HIV-specific prevalence estimates. For some disorders, including substance use disorders, PTSD and insomnia, rates were about 10 times lower in the chart compared to prior literature. Of the disorders sampled, MDD was the only disorder that was documented at expected rates. These findings suggest a potential bias in how clinicians either detect or document psychiatric disorders in PLWH.

In epidemiological studies, rates of psychiatric disorders are consistently higher among PLWH compared to the general population (1–2). In contrast, we found evidence for lower reporting of many psychiatric disorders among PLWH compared to the general population. While clinicians are appropriately documenting evidence of MDD, other psychiatric comorbidities are being largely overlooked.

Sufficient documentation of MDD was somewhat surprising in light of under-documentation for all other psychiatric disorders. However, appropriate documentation of depression among PLWH may be a relatively new phenomenon, as older reports found lower rates of depression documentation among individuals with AIDS compared to individuals without AIDS (19). This finding suggests that clinicians are increasing detection and documentation of depression among PLWH, perhaps because of improved awareness of effective treatments for MDD, including CBT. These findings also raise the possibility that clinicians recognize the psychological impact of HIV-infection, but tend to be biased in their detection and documentation of MDD to the exclusion of other psychiatric comorbidities. This may be due to a lack of recognition of the effectiveness of other EBPs for other disorders, In the general population, primary care physicians tend to over-diagnose MDD, especially among patients with lower psychosocial functioning (20–21). Furthermore, among PLWH, scores of measures of MDD may be inflated due to overlapping somatic symptoms between depression and HIV-infection (22). To improve documentation across psychiatric conditions, efforts should focus on educating clinicians about the complex psychological impact of HIV-infection⁴ as well as on psychiatric predictors of acquiring HIV-infection (23–24).

The consequences of under-documentation of psychiatric disorders among PLWH are enormous. PLWH are unlikely to receive referrals for mental health treatment if their psychiatric illnesses are not documented. This is of major importance, because PLWH already receive suboptimal rates of mental health care (25). When untreated, psychiatric illnesses such as PTSD are unlikely to spontaneously remit (15). Furthermore, psychiatric comorbidities contribute to poor adherence to antiretroviral therapies (5–6, 26), and are therefore less likely to achieve virologic suppression (6,27). Psychiatric comorbidities also contribute to early mortality among PLWH (28); for instance, among individuals who have achieved immunovirologic success, suicide is the second leading cause of death (29). Therefore, under-documentation of psychiatric illness among PLWH is an overlooked challenge to achieving 90-90-90 targets (30). This is an issue in need of immediate correction. One corrective action that could improve detection and documentation of psychiatric illness about PLWH is to mandate continuing education in EBPs for psychiatric disorders for clinicians treating PLWH.

One limitation is that we did not extract a random sample of non-HIV-infected comparison cases. Therefore, we are unable to determine whether the findings are of relevance to EMR documentation more broadly versus under-documentation for PLWH specifically. Future research should directly compare EMR documentation of PLWH to non-HIV-infected comparisons. Another limitation is that all participants were involved in ongoing studies through the Penn Center for AIDS Research (CFAR). Participants in this particular cohort not may be unrepresentative of the entire population of HIV-infected individuals treated through Penn or of the HIV population generally. However, participation in the Penn CFAR studies may provide improved access to care, which might lead to increased documentation of comorbidities. Therefore, documentation rates in this sample may be higher compared to the general population of HIV-infected individuals.

CONCLUSIONS

In summary, this is the first study of its kind to demonstrate remarkable underreporting of psychiatric conditions in EMRs of PLWH. Receipt of psychiatric treatment is critical to the 90-90-90 targets. In order for PLWH to receive psychiatric treatment, they must be properly diagnosed with any relevant psychiatric conditions. These findings indicate a need for strategies to enhance psychiatric diagnosis detection and documentation among PLWH.