Identifying and Characterizing Cancer Survivors in the US Primary Care Safety Net

Abstract

Background

Primary care providers must understand the utilization patterns, clinical complexity, and primary care needs of cancer survivors to provide quality healthcare services. Yet, little is known about the prevalence and healthcare needs of this growing population, particularly in safety net settings.

Methods

We identified adults with a history of cancer documented in primary care electronic health records within a network of community health centers (CHC) in 19 states. We estimated cancer history prevalence among >1.2 million patients and compared sex-specific site distributions to national estimates. Each survivor was matched to three patients without cancer from the same set of clinics. We then compared demographic characteristics, primary care utilization, and comorbidity burden between the two groups, assessing differences with absolute standardized mean differences (ASMD). ASMD values >0.1 denote meaningful differences between groups. Generalized estimating equations yielded adjusted odds ratios (aOR) for select utilization indicators.

Results

We identified 40,266 cancer survivors (prevalence=3.0% of adult CHC patients). Compared to matched cancer-free patients, a higher percentage of survivors had ≥6 primary care visits across 3 years (62% vs 48%) and were insured (83% vs 74%; ASMD>0.1 for both). Cancer survivors had excess medical complexity, including higher prevalence of depression, asthma/COPD, and liver disease (ASMD>0.1 for all). Survivors had higher odds of any opioid prescription (aOR=1.23, 95% CI=1.19–1.27) and chronic opioid therapy (aOR=1.27, 95% CI=1.23–1.32), relative to matched controls (p<0.001 for all).

Conclusions

Identifying cancer survivors and understanding their utilization patterns and physical and mental comorbidities presents an opportunity to tailor primary healthcare services to this population.

Precis:

This large observational study using primary care electronic health record data reveals several important findings about adult cancer survivors, including substantial differences in primary care utilization and health insurance patterns, excess comorbidity, and disparities in chronic pain and opioid use, compared to matched cancer-free counterparts. As more patients continue to survive cancer, the role for primary care providers in providing long-term, comprehensive follow-up care is increasing.

Background

There is a growing population of cancer survivors in the United States: recent estimates put the number of individuals living with a history of cancer at over 15 million,¹ which is expected to double by 2050.² Patients with cancer often lose contact with their primary care clinic when diagnosed and through the duration of treatment and initial follow-up, which typically lasts around 5 years.^2,3 After this period, cancer survivors receive most of their ongoing healthcare services from primary care providers (PCPs).^2,4 For example, a study found only one-third of cancer survivors received any cancer specialty care five years after survival; the majority received all follow-up and routine healthcare services from PCPs.⁴ Cancer survivors often need frequent cancer screening and ongoing management of treatment-related side effects.^2,5 Yet, PCPs often lack adequate information to appropriately address the ongoing needs of this population. Survivorship care plans have been proposed as a strategy to bridge oncology and primary care but their efficacy is still uncertain and they are not yet standard practice.⁶ Even when both PCPs and specialist providers use electronic health records (EHRs), technological barriers persist and communication may not be seamless or complete.⁷

Community health centers (CHCs) and other community-based primary care settings not connected to a hospital, academic institution, or managed care organization, may find it especially difficult to access cancer diagnosis and treatment information.^8,9 Further, individuals receiving care at CHCs and other underserved populations are less likely to receive appropriate cancer screening, are at higher risk for delayed diagnosis, and are more likely to experience preventable morbidity and mortality, particularly for cancer sites associated with modifiable risk factors (e.g., tobacco use is linked to lung cancer).^10–12

‘Safety net’ health care providers deliver essential services regardless of health insurance status.¹³ CHCs represent the US primary care safety net system, receiving federal funding to serve vulnerable populations (i.e., low-income and publicly insured or uninsured patients). Disparities in behavioral cancer risk factors (e.g., tobacco use) and cancer screening persist in this disadvantaged patient population.^14,15 Yet, most cancer survivorship studies to date have been conducted using cancer registry data,^16,17 survey data,¹⁸ or within academic health centers.¹⁹ Little is known about the prevalence of cancer survivors among CHC patients or how much cancer-related data are captured in the outpatient EHRs of CHCs.^20,21 It is also unknown how CHC cancer survivors differ from their counterparts without a documented cancer history on demographic factors, healthcare utilization, and chronic disease burden; an improved understanding of important differences would help providers identify and tailor care to this population.

To address these gaps in the literature, we used primary care EHR data to identify cancer survivors and a matched population of patients seen in CHCs without a documented diagnosis of cancer. Specifically, we used EHR data from 431 CHCs in 19 states to estimate the prevalence of cancer history and to describe a range of cancer-related characteristics and comorbidities.

Methods

Data Sources

The Accelerating Data Value Across a National Community Health Center Network (ADVANCE) clinical data research network (CDRN), one of 13 PCORnet CDRNs,²² is a multi-center collaborative led by the OCHIN (formerly Oregon Community Health Information Network but shortened to OCHIN when expanded to other states) network in partnership with Health Choice Network (based in Florida) and Fenway Health (based in Massachusetts).²³ The three member organizations are comprised of CHCs spanning 24 states which serve disadvantaged and vulnerable populations.²⁴ EHR data from each partner organization are integrated and standardized into a common data model. The demographic profile of the ADVANCE patient population mirrors national CHC estimates²⁵ and is generalizable to the US safety net population.

Study Population

We identified 42,359 cancer survivors alive and aged ≥19 as of 12/31/2016. To be included, patients had to have ≥1 office visit at one of 431 primary care clinics that were actively using an EHR system throughout the three-year study period (2014–2016), and had to have a record in a discrete searchable EHR field (i.e., encounter diagnosis, problem list, medical history) indicating a malignant cancer diagnosis, excluding non-melanoma skin cancer. These cancer survivors represented 3.0% of our CHC patient population within the 19 states with active primary care clinics in the ADVANCE network.

To construct a comparison group, we identified >1.2 million adult patients from the same clinics with no documented cancer diagnosis as of 12/31/2016. This population was younger and had a different sex distribution than cancer survivors; thus, to reduce bias and improve group comparisons, cancer survivors were matched in a 1:3 ratio²⁶ to this comparison group. We matched on sex, year of birth, and primary health system (each patient’s most frequently accessed CHC): for each cancer survivor, 3 matches were selected at random and without replacement. This exact matching method resulted in exclusion of 2,093 cancer survivors because they had <3 control matches available (4.9% of survivors). Excluded patients were more commonly male, non-Hispanic white, and older than included cancer patients. The final dataset contained 40,226 cancer survivors with 47,339 primary cancer sites (as a single patient can have >1 cancer site) and 120,798 matched comparison patients.

Variables

Cancer diagnoses were grouped into primary sites following classifications from the US Surveillance, Epidemiology, and End Results (SEER) Program (https://seer.cancer.gov/tools/conversion/). To provide estimates of data completeness and general population representativeness, we compared sex-stratified rankings and prevalence for the most common cancer sites to national estimates.¹ For each survivor, we calculated the number of distinct cancer sites recorded, length of time since diagnosis and age at first diagnosis (based on EHR-reported onset or noted dates), comparing to national estimates where available (Supporting Information, Tables 1–2).

Cancer survivors were compared to matched cancer-free counterparts on race, ethnicity, preferred language, and household income ≤138% of federal poverty level [(FPL) the cut-point for Medicaid eligibility under Affordable Care Act reform]. We also compared the groups on insurance type, length of time established at their primary CHC, whether patients were ever uninsured or consistently uninsured throughout the study period, counts of office visits, number of providers seen, and primary care visits (see Table 2 for definitions).

Table 2.

Healthcare access and utilization, matched CHC patients with and without history of malignant cancer

	Cancer history† N=40,266	No cancer history N=120,798	Absolute standardized mean difference
Insurance Type (at last visit)			0.214
Medicaid	12,606 (31.3%)	32,253 (26.7%)
Medicare	13,376 (33.2%)	33,749 (27.9%)
Private	7,563 (18.8%)	23,931 (19.8%)
Uninsured	5,549 (13.8%)	26,549 (22.0%)
Other/Unknown	1,172 (2.9%)	4,316 (3.6%)
Ever uninsured, 2014–2016	11,747 (29.2%)	41,924 (34.7%)	0.119
Consistently uninsured, 2014–2016	3,875 (9.6%)	21,022 (17.4%)	0.229
Length of time established at primary CHC			0.218
< 1 year	8,711 (21.6%)	39,182 (32.4%)
1-<5 years	19,827 (49.2%)	53,418 (44.2%)
5-<10 years	11,145 (27.7%)	26,951 (22.3%)
>10 years	583 (1.4%)	1,247 (1.0%)
PCP assigned	34,754 (86.3%)	100,369 (83.1%)	0.083
Total office visits, 2014–2016			0.197
<3	7,188 (17.9%)	35,546 (29.4%)
3–5	8,259 (20.5%)	27,659 (22.9%)
6–10	9,974 (24.8%)	26,641 (22.1%)
11–24	11,538 (28.7%)	25,004 (20.7%)
≥25	3,307 (8.2%)	5,948 (4.9%)
PC office visits‡, 2014–2016			0.257
<3	9,450 (23.5%)	43,821 (36.3%)
3–5	9,121 (22.7%)	28,637 (23.7%)
6–10	10,632 (26.4%)	26,200 (21.7%)
11–24	9,616 (23.9%)	19,710 (16.3%)
≥25	1,447 (3.6%)	2,430 (2.0%)
Number of providers§ seen, 2014–2016			0.196
<3	19,664 (48.8%)	71,108 (58.9%)
3–5	14,389 (35.7%)	36,241 (30.0%)
6–9	4,963 (12.3%)	10,955 (9.1%)
≥10	1,250 (3.1%)	2,494 (2.1%)

Note: Cancer survivors were matched in 1:3 ratio with cancer-free patients on sex, year of birth, and primary health center. We considered an absolute standardized mean difference of >0.1 to denote meaningful differences between the groups.

^†Includes all malignant cancer diagnoses except non-melanoma skin cancer.

^‡CPT codes 99201–99205; 99212–99215; 99241–99245; 99381–99387; 99391–99397 with a primary care provider

^§Includes physicians, naturopathic physicians, physician assistants, and nurse practitioners

We compared cancer survivors to their matched cancer-free counterparts on Charlson Comorbidity Index (CCI), pain, and opioid use. An enhanced version of the CCI,²⁷ which considers additional physical, mental, and behavioral health conditions, was applied to the EHR problem list (see Table 3 footnotes). We disregarded cancer diagnoses from our CCI calculation for a more accurate comparison of non-cancer comorbidity burden between the groups. The 10 most prevalent conditions among our study population were presented individually. Chronic pain diagnoses were identified from problem list records, and prescription orders were used to identify those who were prescribed opioid medications. Chronic opioid use was defined as ≥60 tablets or any fentanyl patch prescribed in any 90-day period.²⁸ We identified opioid use disorder using both problem list and encounter diagnoses. All time-varying characteristics (e.g., FPL, CCI) were assigned as of each patient’s last encounter in the study period.

Table 3.

Medical comorbidity, matched CHC patients with and without history of malignant cancer, 2014–2016

	Cancer history† N=40,266	No cancer history N=120,798	Absolute standardized mean difference
Adapted Charlson Comorbidity Index‡			0.217
0–1	17,979 (44.7%)	65,182 (54.0%)
2–3	10,731 (26.7%)	29,582 (24.5%)
4–5	6,335 (15.7%)	15,399 (12.7%)
6–9	4,470 (11.1%)	9,414 (7.8%)
≥10	751 (1.9%)	1,221 (1.0%)
Chronic pain condition§	3,099 (7.7%)	6,857 (5.7%)	0.081
Prescribed any opioid	10,058 (25.0%)	20,281 (16.8%)	0.203
Chronic opioid therapy¶	7,234 (18.0%)	13,666 (11.3%)	0.189
Opioid use disorder#	542 (1.2%)	1,107 (0.8%)	0.041
Individual conditions (in order of prevalence, components of Charlson Comorbidity Index)
Hypertension	19,316 (48.0%)	53,595 (44.4%)	0.072
Depression	9,650 (24.0%)	22,492 (18.6%)	0.132
Diabetes	8,404 (20.9%)	23,316 (19.3%)	0.039
COPD and asthma	7,761 (19.3%)	17,259 (14.3%)	0.134
Drug or alcohol abuse	7,020 (17.4%)	16,706 (13.8%)	0.099
Liver disease	2,850 (7.1%)	5,619 (4.7%)	0.103
Renal disease	2,541 (6.3%)	5,543 (4.6%)	0.076
Cerebrovascular disease	2,220 (5.5%)	5,316 (4.4%)	0.051
Bipolar disorder	1,403 (3.5%)	3,215 (2.7%)	0.048
Congestive heart failure	1,341 (3.3%)	3,289 (2.7%)	0.036

Note: Cancer survivors were matched in 1:3 ratio with cancer-free patients on sex, year of birth, and primary health center. We considered an absolute standardized mean difference of >0.1 to denote meaningful differences between the groups.

^†Includes all malignant cancer diagnoses except non-melanoma skin cancer.

^‡Charlson comorbidity index excluding cancer component, calculated from active problem list diagnoses as of 12/31/2016; ‘enhanced’ index includes added weights for transplant history, IBD, seizures, sickle cell anemia, hemophilia, muscular dystrophy, Down syndrome, cystic fibrosis, Tay-Sachs disease, developmental delay, mental retardation, cerebral palsy, autism, schizophrenia, bipolar disorder, and drug or alcohol abuse. Used by permission (Mary Charlson, 2017)

^§ICD9: 338.*, 729.1–729.2, 780.71; ICD10: G89.2-G89.4, M54.10, M54.18, M60.8-M60.9, M79.1, M79.2, M79.7, and R53.82

^¶≥60 opioid tablets or fentanyl patch prescribed in any 90-day period

^#ICD9: 304.00, 304.01, 304.02; ICD10: F11.1x or F11.2x

Analysis

This cross-sectional analysis utilized data across a three-year study period (2014–2016). In univariable analyses, we present patient demographics, healthcare utilization, and prevalence of chronic conditions by cancer history groups. Given the large sample size, we assess differences between the groups using absolute standardized mean differences (ASMD), which are not affected by sample size. ASMD is an effect size measure increasingly used in observational studies to compare distributional differences between groups, with extensions to binary and multinomial variables.²⁹ ASMD is defined as the difference in group means in units of standard deviation,³⁰ and ranges from 0 (indicating that the groups are equivalent on the measure being compared) to 1.0 (indicating perfect disagreement). As has been done in other work, we considered an ASMD of >0.1 (interpreted as 10% difference) to denote meaningful differences between the groups.³⁰

In multivariable analyses, we compared four utilization indicators for cancer survivors relative to their cancer-free counterparts controlling for important confounders through regression modeling. To account for the correlated data arising from the relatedness of cancer survivors and their cancer-free counterparts through matching, we used generalized estimating equation (GEE) logistic regression models, with standard errors clustered on match set to test for differences between cancer survivors and cancer-free counterparts.³¹ We computed adjusted odds ratios (aOR) with 95% confidence intervals (CI) adjusted for race/ethnicity, CCI, insurance type, years established, PCP assignment, and number of office visits. Statistical significance was two-sided and set at α=0.05.

Due to the clinical heterogeneity of cancer and its treatment, we also present demographic, utilization, and clinical measures for survivors stratified by the leading five sites for men and women (Supporting Information, Table 3). Data management and analysis were conducted using SAS software version 9.4 (SAS Institute, Inc., Cary, NC, USA). The Oregon Health & Science University Institutional Review Board approved this study.

Results

The leading 10 cancer sites in this CHC population were the same as leading sites reported nationally,¹ however some rankings differed (Supporting Information, Table 1). Most cancer survivors were female (62.1%), with a median age of 60. Compared to sex- and age-matched cancer-free counterparts from the same CHCs, a higher proportion of survivors were white (60.5% vs 53.9%), whereas a lower percentage were Hispanic (21.3% vs 25.5%) and spoke a primary language other than English (17.0% vs 22.7%; Table 1). The two groups had similar distributions of household income (~62% ≤138% FPL).

Table 1.

Demographic characteristics of matched CHC patients with and without history of malignant cancer

	Cancer history†N=40,266	No cancer history N=120,798	Absolute standardized mean difference
Sex			--
Female	25,016 (62.1%)	75,048 (62.1%)
Male	15,243 (37.9%)	45,729 (37.9%)
Age (as of 12/31/2016)			--
19–29	1,206 (3.0%)	3,618 (3.0%)
30–39	2,587 (6.4%)	7,761 (6.4%)
40–49	4,711 (11.7%)	14,133 (11.7%)
50–59	10,552 (26.2%)	31,657 (26.2%)
60–69	12,645 (31.4%)	37,934 (31.4%)
70–79	6,138 (15.2%)	18,414 (15.2%)
≥80	2,427 (6.0%)	7,281 (6.0%)
Race/ethnicity			0.220
Hispanic	8,573 (21.3%)	30,806 (25.5%)
Non-Hispanic White	24,347 (60.5%)	65,151 (53.9%)
Non-Hispanic Black	5,305 (13.2%)	16,716 (13.8%)
Non-Hispanic Asian	734 (1.8%)	3,557 (2.9%)
Non-Hispanic Other	661 (1.6%)	2,160 (1.8%)
Unknown	646 (1.6%)	2,408 (2.0%)
Primary language			0.194
English	33,425 (83.0%)	93,346 (77.3%)
Spanish	5,612 (13.9%)	21,828 (18.1%)
Other/Unknown	1,229 (3.1%)	5,624 (4.7%)
FPL (latest recorded)			0.026
≤138%	24,834 (61.7%)	74,251 (61.5%)
≥139%	7,670 (19.0%)	22,574 (18.7%)
Unknown	7,762 (19.3%)	23,973 (19.8%)

Note: Cancer survivors matched in 1:3 ratio with cancer-free patients on sex, year of birth, and primary health center. We considered an absolute standardized mean difference of >0.1 to denote meaningful differences between the groups.

^†Includes all malignant cancer diagnoses except non-melanoma skin cancer.

Compared to their matched cancer-free counterparts (Table 2), CHC cancer survivors were more often publicly insured (Medicaid or Medicare: 64.5% vs 54.6%) and had a lower prevalence of being uninsured at their last visit (13.8% vs 22.0%) or ever uninsured in the study period (29.2% vs 34.7%). Cancer survivors had been established longer at their primary CHCs, and utilized more healthcare services, in terms of the number of office visits, primary care visits, and providers seen. The largest differences in magnitude were seen when comparing the number of primary care visits (ASMD=0.257) and total office visits (ASMD=0.197) between survivors and matched controls.

CHC cancer survivors had much higher rates of medical complexity and chronic pain than cancer-free counterparts. Even when disregarding their cancer history, 28.7% of survivors had a CCI ≥4, versus 21.6% of matched controls (ASMD=0.217, Table 3). They also had a higher prevalence of chronic pain conditions (7.7% vs 5.7%, ASMD=0.081), having any opioid prescribed in the study period (25.0% vs 16.8%, ASMD=0.203), and using opioids at a chronic level (18.0% vs 11.3%, ASMD=0.189). Additionally, cancer survivors had a greater burden of each individual chronic condition assessed compared to cancer-free counterparts: 48.0% of cancer survivors had hypertension (vs 44.4%, ASMD=0.072) and 23.9% had depression (vs 18.6%, ASMD=0.132). Comorbidity burden differed by primary site; for example, lung cancer survivors had the highest CCI scores and prevalence of opioid use, while prostate cancer survivors were more likely to have hypertension and renal disease and less likely to have a depression diagnosis than survivors of other cancers (Supporting Information, Table 3).

After adjusting for all covariates, cancer survivors had 45% increased odds of having at least 6 office visits in the three-year study period (aOR 1.45, 95% CI: 1.41–1.49; Figure 1), 23% increased odds of any opioid use (aOR 1.23, 95% CI: 1.19–1.27), and 27% increased odds of chronic opioid therapy (aOR 1.27, 95% CI: 1.23–1.32) relative to matched cancer-free counterparts. Survivors and comparison patients had similar odds of PCP assignment.

Figure 1. Adjusted odds ratios for select utilization indicators, CHC cancer survivors vs matched patients with no cancer history.

Legend: Cancer survivors include those with any malignant cancer diagnosis excluding non-melanoma skin cancer. Cancer survivors were matched in 1:3 ratio with cancer-free patients on sex, year of birth, and primary health center. Odds ratios obtained from covariate-adjusted GEE models with standard errors clustered on match set ID to account for within-matched set correlation. All models controlled for race/ethnicity, adapted Charlson Comorbidity Index, insurance type, any uninsured visit in the study period, and years established with the primary health system; PCP model additionally adjusted for office visit count; visit utilization model additionally adjusted for PCP assignment; any opioid and chronic opioid models additionally adjusted for office visit count and PCP assignment. PCP model: p=0.03; all other models: p<0.001.

Discussion

We found 3.0% of adult patients in this CHC population had a documented history of cancer in the primary care EHR. This percentage is lower than expected, as nationally about 6.7% of the US adult population are estimated to be survivors of non-skin cancers;³² however, CHC populations are younger than the overall US population,²⁵ likely explaining the lower prevalence of survivor status. In addition to a younger population, CHC patients are poorer, more commonly racial-ethnic minorities,²⁴ and face excess barriers to screening³³ than the general population, all characteristics that align with demographic differences in cancer diagnosis and survival rates.^34–36 The 10 leading cancer sites that we found in this population were the same as those seen in the general US population for top 10 cancer sites, although we found somewhat different ranks and prevalence estimates. For example, breast cancer was the most common site for women both in the CHC sample and nationally, but our EHR data suggest that only 32% of female survivors had this type of cancer versus 44% in national data¹ (see Supporting Information, Table 1).

A growing body of literature has documented the excess chronic disease burden and associated symptoms experienced by cancer survivors.³⁷ For example, up to 40% of cancer survivors report suffering from chronic pain,^38,39 which may persist a decade or more after treatment. Our finding that cancer survivors had about 25% higher odds of opioid use than their matched cancer-free counterparts adds to this literature and points to a topic for future inquiry. Within a few years after diagnosis, the management of chronic pain and other potential latent effects of cancer and its treatment is likely addressed in the primary care setting^2,4. While a fair amount of attention has been given to strengthening the transition from oncology to primary care,² more work is needed on how best to transition cancer information and pain management therapy, including opioids, between care providers and settings.⁴⁰ Subgroup examinations suggest that survivors of certain cancers (e.g., lung and bronchus, non-Hodgkin lymphoma, and colorectal) have higher rates of opioid prescribing and chronic opioid therapy (see Supporting Information Table 3); further study of cancer survivorship and opioid use in the primary care setting could lead to important interventions.

Our finding that cancer survivors suffer disproportionately from multiple chronic conditions including diabetes, hypertension, and mental health conditions, further speaks to the need for care coordination and attention to secondary preventive care. Cancer survivors in this study had lower rates of uninsurance and higher prevalence of being publicly insured than their cancer-free counterparts; yet, over one-quarter had at least one period without health insurance and nearly 14% were consistently uninsured in the three-year study period. Even among CHC patients who can access care regardless of insurance status, patients are less likely to receive preventive care while uninsured.^41,42 Any instability in health insurance coverage could be especially detrimental to this high-need population.⁴³

Survivorship care plans may help bridge gaps in care during the transition from oncology to primary care,^2,44 and could feasibly be exchanged across EHRs.⁴⁵ To date, however, this objective has yet to be fully realized. In integrated healthcare settings and pilot studies, EHRs have facilitated coordination and transition of care between oncologists and PCPs,^9,46 but little evidence of this process has been documented in other settings, such as CHCs, or across different EHR systems. Low levels of adoption of survivorship care plans within oncology⁴⁷ has the trickle-down effect of poor care coordination and incomplete follow-up guidelines reaching PCPs.⁴⁶The development of standard flowsheets or templates for survivorship care plans in EHRs may facilitate systematic study of survivorship care plans and address the lack of definitive data showing patient benefits of these widely-encouraged tools.⁷

Our study has several limitations. The CHCs in our study are not organizationally affiliated with cancer centers, academic research institutions, or integrated health systems, thus ascertainment of cancer history is likely reliant on patient self-report and may be incomplete. Further, we were limited to discrete searchable EHR fields to identify cancer history; more complete ascertainment may be achieved through manual chart review but this was beyond the scope of our study. Our comparisons to national estimates reveal some apparent discrepancies in primary cancer site prevalence. This may be explained by lack of specificity in primary site codes in the outpatient EHR, compared to cancer registry data on which national estimates are based. Data on cancer survival, stage and other diagnostic characteristics are also lacking in the EHR. Linkages to central cancer registries could help validate and improve the EHR-recorded data on cancer history and determine the extent to which cancer survivorship status is under-captured in primary care EHRs. A final limitation is the exclusion of a small percent of cancer survivors from analysis due to not having three matched controls. Despite these limitations, we successfully identified a large cohort of cancer survivors accessing primary care in safety-net clinics and documented many important differences for consideration in the provision of primary care.

As more patients continue to survive cancer, the role for primary care teams in providing comprehensive follow-up care is increasing.⁴⁸ Taken together, our results can inform primary care practice. First, EHRs allow the opportunity to systematically identify cancer survivors in primary care settings, which is the first step toward effective care coordination. Second, understanding the primary care utilization patterns and physical and mental comorbidity burden of cancer survivors presents an opportunity to prioritize their healthcare needs in CHCs’ population health efforts. Third, particular focus should be paid to the chronic pain management needs of this population to reduce risks associated with opioid use and mitigate disparities in substance misuse. EHR-based cancer history information could conceivably be leveraged to develop clinical decision support tools, prompting primary care providers to tailor care to the specific needs of this growing population.