Patterns of Pulmonary Consultation for Veterans with Incident Chronic Obstructive Pulmonary Disease

To the Editor:

Chronic obstructive pulmonary disease (COPD) is a common, debilitating condition. Although COPD is usually managed by primary care providers (PCPs), pulmonologists are often consulted to assist in complex disease management (1). Yet, there is no consensus about when to refer patients for specialist care (1 –3).

The decision to involve pulmonologists must weigh the potential benefits of specialist expertise against the challenges of coordinating care across multiple providers. Successful care coordination depends on developing clear roles and responsibilities across the “specialty care triad,” which includes PCPs, specialists, and patients (Figure 1) (4, 5). Understanding how characteristics of these groups influence consultation for incident COPD may help guide successful use of specialty services. We hypothesized that patients with multiple comorbidities, providers with less experience and larger panels, and facilities with an on-site pulmonologist would be more likely to consult pulmonology for incident COPD.

Figure 1.

Specialty care triad of patient, provider and facility characteristics that influence consultations to pulmonary specialist for COPD. The above listed covariables were included in the multivariable logistic model (Table 2) based on the specialty care conceptual model. COPD = chronic obstructive pulmonary disease.

A portion of the work contained in this manuscript has been previously published in abstract form (6).

Methods

We conducted a retrospective cohort study of pulmonary consultations for veterans with an incident COPD diagnosis (based on diagnostic codes not previously documented in Veterans Health Administration (VA) or in U.S. Centers for Medicare and Medicaid Services [CMS]) during fiscal years 2010–2015. We excluded veterans who had pulmonary visits in VA or CMS during the prior year and those younger than 65 years. We assigned a VA PCP based on the provider with the most primary care visits in the same quarter as the incident COPD diagnosis and assigned facility based on PCP practice location.

We collected data on veteran demographics, comorbidities, and Elixhauser score (7 –9). We also collected PCP and facility characteristics. We categorized variables that had nonlinear associations with pulmonary referrals.

Our primary outcome was outpatient pulmonary clinic visit in the year after incident COPD diagnosis. We were not able to identify requested pulmonary referrals that did not result in pulmonary encounters. We conducted a multilevel logistic regression analysis, including both provider and site variables as random effects.

Results

We identified 180,044 veterans with incident COPD. Overall, 31,379 (17.4%) veterans had a pulmonary visit within one year, with visits generally increasing over time (from 15.8% in 2010 to 18.9% in 2014). Identified veterans tended to be older (mean 73.9 yr), white (81.7%) males (98.6%) with frequent comorbidities (Table 1). We identified 11,959 PCPs, who were 43% female and 44% nonphysicians (missing sex for 2,558 and physician status for 1,057). Mean PCP panel size was 881 patients and 46.7% of PCPs had less than 1 year of VA tenure. Of the facilities where veterans received their care, 20.6% were academically affiliated, 96.4% had on-site pulmonary specialists, and 9.7% were rural.

Table 1.

Baseline characteristics of veterans with incident chronic obstructive pulmonary disease with and without pulmonary visits

Patient Characteristics	Pulmonary Visit (n = 31,379) (%)	No Pulmonary Visit (n = 148,665) (%)
Age, yr
65–69	40.6	40.2
70–74	19.0	18.9
75–79	16.2	14.9
80–84	13.2	13.1
≥85	11.0	12.9
Sex, female	1.4	1.4
Race
White	80.0	81.5
Black	9.1	7.4
Hispanic	2.6	2.4
Other	1.9	2.0
Unknown	6.5	6.7
Hospitalization in prior year
0	60.2	67.8
≥1	39.8	32.2
Heart failure	16.1	13.4
Arrhythmia	23.4	20.3
Valvular heart disease	6.3	5.1
Pulmonary hypertension	5.5	3.5
Peripheral vascular disease	14.0	13.9
Malignancy	17.0	15.8
Obesity	19.2	17.3
Substance use disorder	8.2	9.5
Depression	22.3	22.7
Anxiety	4.6	4.6

In multivariable regression, several patient, PCP, and site characteristics were associated with pulmonary visits (Table 2). Older and minority veterans were less likely to have pulmonary consults, including Black and Hispanic veterans. Pulmonary consultations were also less likely among patients with substance use disorder and depression. In contrast, pulmonary consultations were more likely in veterans with prior hospitalizations, higher Elixhauser scores, cardiovascular comorbidities, and obesity.

Table 2.

Multivariable regression of patient, PCP, and site characteristics associated with pulmonary consultation among veterans with incident COPD

Covariates		n		OR*		95% CI
Patient characteristics
Age, yr
65–69		72,569		Ref		—
70–74		33,996		0.97		0.93–1.00
75–79		27,275		1.01		0.97–1.05
80–84		23,625		0.90		0.87–0.94
≥85		22,579		0.73		0.69–0.76
Sex, female		2,535		0.99		0.89–1.11
Race
White		146,245		Ref		—
Black		13,830		0.87		0.83–0.92
Hispanic		4,326		0.88		0.80–0.97
Other		3,616		0.98		0.89–1.07
Hospitalization in prior year
0		119,622		Ref		—
≥1		60,422		1.27		1.23–1.31
Heart failure		24,946		1.01		0.97–1.05
Arrhythmia		37,596		1.07		1.03–1.10
Valvular heart disease		9,514		1.06		0.997–1.12
Pulmonary hypertension		6,863		1.33		1.25–1.41
Peripheral vascular disease		25,052		0.90		0.87–0.94
Malignancy		28,765		1.01		0.97–1.04
Obesity		31,740		1.08		1.04–1.12
Substance use disorder		16,669		0.77		0.73–0.80
Depression		40,747		0.94		0.91–0.97
Anxiety		8,307		0.98		0.92–1.04
Elixhauser Comorbidity Index		180,044		1.03		1.02–1.04
Provider characteristics
VA tenure, yr
<1		66,996		Ref
1–5		35,902		1.10		1.05–1.15
6–10		35,736		1.11		1.06–1.17
>10		41,410		1.20		1.14–1.25
Professional role
Physician		131,342		Ref
Nonphysician provider		39,196		0.97		0.93–1.01
Unknown		9,506		0.92		0.84–0.99
Sex of PCP
Female		74,201		Ref
Male		71,548		1.04		1.01–1.09
Unknown		34,295		0.97		0.92–1.02
Panel size
<500		11,343		Ref
500–1,000		52,578		1.09		1.02–1.16
1,000–1,500		102,899		1.06		0.996–1.13
>1,500		13,224		1.04		0.95–1.14
Site characteristics
Pulmonary clinic on-site		173,583		1.61		1.21–2.14
Academic		37,083		2.58		2.46–2.70
Rural		17,554		0.87		0.72–1.06

Definition of abbreviations: CI = confidence interval; COPD = chronic obstructive pulmonary disease; OR = odds ratio; PCP = primary care provider; Ref = reference; VA = Veterans Health Administration.

All covariates listed in the table were included in the multivariate model, and there are no other variables that are not listed.

Variables with missing data: sex (n = 2,558) and physician status (n = 1,057).

^*Bold typeface ORs are significant at P < 0.05.

PCPs with longer tenures or larger panel sizes tended to be more likely to consult pulmonary. Veterans at academic facilities and facilities with available pulmonary clinics were also more likely to have pulmonary visits.

Discussion

In this large, longitudinal study of veterans with incident COPD, we identified several key findings that expand knowledge of the specialty care triad for COPD and identify characteristics associated with pulmonary consultation. We found that 17.4% of veterans with incident COPD had a subsequent pulmonary visit, which is higher than the 10.7% of patients referred in a Canadian study (10). This may reflect differences in health systems and patient characteristics as veterans tend to have complex multimorbidity. Our findings confirmed our hypothesis that pulmonary consults were more common among veterans with prior hospitalizations, higher Elixhauser scores, and multiple comorbidities, reflecting an appropriate use of specialty care (11 –13).

We found that pulmonary consultations were less common among veterans who frequently encounter health disparities, including Black individuals, Hispanic individuals, and those with substance use and depression, which may reflect systemic bias in deciding which patients warrant pulmonary consultation. Previous studies have demonstrated notable racial and sex disparities in pulmonary disease (14 –16), and mental health disorders may result in barriers to care (17 –19). Our study suggests that these disparities also extend to the process of pulmonary consultation. Potential opportunities to support specialty care delivery could include evolving technologies such as telehealth or policy implementation like VA MISSION Act (20, 21).

We found wide variation in the tendency to obtain pulmonary consultations across PCPs, which may relate to their experience and workload. Supporting our hypothesis, we found that PCPs with larger patient panels were more likely to obtain pulmonary consultation. Pressures to increase PCP productivity ought to consider unintended consequences of dispersing care across multiple providers and potentially increasing care fragmentation. However, contrary to our hypothesis, we found that PCPs with more experience were more likely to obtain pulmonary consults, which may reflect more complex patient panels.

PCP practice setting was also associated with pulmonary consultations, confirming our hypothesis and consistent with prior work showing that academic facilities and those with pulmonary clinics had more pulmonary visits (10, 22). This finding highlights the importance of supplying equitable distribution of specialists within healthcare systems (23).

This study has limitations. Administrative data could misclassify veterans with COPD, and we did not have disease severity information on spirometry, oxygen, or noninvasive ventilation use (24). Our findings may not generalize to other healthcare systems. We were unable to discern whether patients received a referral for pulmonary consultation that they did not attend. Lastly, our analyses are based on the specialty care triad conceptual framework, which does not identify any relationships between predictor variables. It is possible that there are unaccounted relationships in our model, and future exploratory research should focus on mechanistic discovery of mediators and interactions between predictors influencing pulmonary consults.

In conclusion, our national study expands the limited research on pulmonary consultations for COPD. We found higher rates of pulmonary consultation than previously described, with visits increasing over time. Unexpectedly, we identified health disparities in pulmonary consultations for racial and ethnic minorities as well as veterans with substance use and mental health disorders. Finally, we found PCP and site characteristics that suggest systemic factors influencing pulmonary consultation. These findings highlight the consequences of practice environments that could contribute to fragmented care and health inequalities. Our findings serve as a call for further research to identify appropriateness of pulmonary consultations and develop guidelines that address disparities and promote equity.