The Affordable Care Act’s Medicaid Expansion and Impact Along the Cancer-Care Continuum: A Systematic Review

Conceptual Framework

This systematic review adheres to the guidelines set by the Preferred Reporting Issues for Systematic Review and Meta-Analysis standards for systematic review of studies that evaluate health-care interventions and uses the Covidence platform (an online Cochrane primary screening and data extraction tool) to ensure a rigorous methodology and reporting (13, 14).

Search Strategy

A medical librarian (S. Kaplan) searched MEDLINE (via PubMed), Embase (via Elsevier), Scopus (via Elsevier), and the Cochrane Registry of Trials (via Wiley) using a combination of keywords and database-specific controlled vocabulary terms for the concepts: Medicaid Expansion and Cancer (available online) (Appendix 1). Because the review concerned changes resulting from the ACA—American legislation passed in 2010 but largely not in effect until 2014—results were restricted to publications published December 1, 2013, to January 29, 2020. To evaluate the impact of Medicaid expansion on changes in health behaviors that could potentially reduce the risk of cancer, such as smoking cessation, we conducted a manual search for relevant publications. Articles were further cross-referenced to identify additional articles. All citations were compiled into EndNote and then imported into Covidence.

Study Selection

We included original peer-reviewed research reporting quantitative, quasi-experimental, or mixed-methods results regarding the exposure of Medicaid expansion on outcomes along the cancer-care continuum (prevention, screening, diagnosis, treatment, survivorship, and end-of-life care) (Figure 2) (12). We defined Medicaid expansion as states that expanded Medicaid for patients with a household income 138% above the federal poverty level from 2010 and onward. We excluded studies that examined the effects of Medicaid expansion before or not part of the ACA, such as the Oregon Health Insurance Experiment. Review articles, published abstracts, conference proceedings, and forecast analyses were excluded. Finally, we excluded articles pertaining to legal, political, or ethical aspects of the ACA, including letters to the editor, policy briefs, and governmental reports.

Titles and abstracts were reviewed for eligibility independently by 2 investigators (H. Moss and J. Wu). Any disagreements were resolved by discussion with a third author (Y. Zafar). The full texts of the eligible articles were uploaded into Covidence, which were subsequently reviewed (HM and JW) to ensure the study met the predefined inclusion and exclusion criteria as noted above.

Data Extraction

A standardized data abstraction form was used to extract data on sample characteristics, study design, and relevant outcomes. Data extraction was performed by 2 investigators (HM and JW), and differences were resolved by discussion and consensus. Outcomes of interest included access to insurance coverage, timeliness of care, incidence/survival metrics, changes in care affordability, and impact on racial and socioeconomic disparities. Outcomes were not mutually exclusive and can occur across the cancer-care continuum.

Two independent investigators (HM and JW) applied the Risk of Bias in Non-randomized Studies of Interventions (ROBINS-I) tool to assess bias (Appendices 2 and 3, available online) and the National Institutes of Health (NIH) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies to assess quality (Appendix 4, available online) for each study. ROBINS-I assesses bias across 7 domains (confounding, participant selection, intervention classification, departure from intended intervention, missing data, measurement of outcomes, and selection of reported results), and the risk of bias is then determined through a combination of all the domains (15). The NIH Quality Assessment Tool includes a 14-item checklist to evaluate nonrandomized studies by considering study objectives, timing between exposure and outcomes, and risk of confounders. Items 4, 10, and 12 were omitted because they were not applicable to these observational studies. Each study was rated as “good,” “fair,” and “poor” based on answers to the tool questions (16). Discrepancies were resolved by consensus. All studies meeting the inclusion criteria were included in the analysis regardless of rating.

Eligible Studies

According to the search strategy, 187 records were identified excluding duplicates, of which 128 did not meet inclusion criteria. A total of 59 records were screened for eligibility by full text review. An additional 11 articles were excluded for not meeting the eligibility criteria. Nine additional articles were included after cross-reference to capture cancer risk–modifying behaviors that did not include “cancer” in the title or abstract. A total of 48 studies were included in this review. Figure 3 provides a Preferred Reporting Issues for Systematic Review and Meta-Analysis flowchart detailing study identification, screening, eligibility, and selection process. Study characteristics are summarized in Tables 1 and 2. Public and government surveys were the most common data source (47%), followed by cancer registries (33%). Approximately 43% of studies assessed data just 1-2 years after Medicaid expansion, 35% for 2-3 years, and 19% assessed 3 years or more. The majority (approximately 90%) of the studies evaluated more than 1 expansion state. In the 48 studies, 124 outcomes were evaluated, and the authors noted that these outcomes are not mutually exclusive and overlap can occur across the cancer-care continuum. The most common outcomes assessed were the impact of expansion on insurance coverage (23%), followed by evaluation of disparities (18%) and access to screening (15%). Other outcomes assessed include prevention (9%), stage at diagnosis (9%), incidence (4%), access to treatment (4%), time to treatment (2%), survival (3%), care affordability (6%), and other health and quality of care (8%).

Several methods used in the studies included evaluation of states pre- and post-ACA implementation, comparisons of expansion states with nonexpansion states, and focus on lower socioeconomic populations and racial and ethnic minorities. The most commonly used analytic method was a difference-in-differences analysis (n = 29). This approach allows researchers to evaluate the impact of a policy while accounting for trends that may exist in both the exposure and comparison groups (65). The included cancer diagnoses, data sources, and number of expansion vs nonexpansion states are further detailed in Table 1.

Study Participant Characteristics

Along the cancer-care continuum, 11 studies examined the impact of expansion during the prevention stage and 18 studies during screening. Nineteen studies focused on outcomes at the time of diagnosis, 6 evaluated patients at the time of treatment, and 6 evaluated outcomes during survivorship. No studies examined outcomes at the end of life (Figure 2). There was heterogeneity in the cancer types, data sources, methods, and number of expansion states analyzed in each stage of the cancer continuum. Most studies evaluated nonelderly adults between ages 18 and 64 years. Six studies included patients younger than 18 years (17, 21, 27, 28, 31, 35). For the prevention stage, 10 studies examined the impact of expansion on smoking cessation. For the screening stage, 10 studies used the Behavioral Risk Factor Surveillance System (BRFSS) database, with most studies examining a combination of breast, cervical, and colorectal cancer. For the diagnosis stage, 10 studies used the National Cancer Institute’s Surveillance, Epidemiology, and End Results Program (SEER) database, 4 studies used the National Cancer Data Base, and 9 studies examined all types of cancer. For the treatment stage, 2 studies used the Healthcare and Utilization Project clinical dataset using various cancer types. For the survivorship stage, 5 studies examined all types of cancer diagnoses.

Quality and Bias Assessment

Using the NIH-tool measure, study quality was rated as “good” for 33 studies, “fair” for 13 studies, and “poor” for 2 studies. The majority of studies with higher scores used national cancer registries and followed up patients for multiple years following expansion. Studies with lower scores evaluated outcomes that required longitudinal assessment (such as stage shift or changes in treatment patterns) but evaluated the impact of only 1 year of expansion. The overall risk of bias was moderate for all studies based on the ROBINS-I tool (Table 1).

Prevention

Eleven studies focused on the impact of Medicaid expansion on cancer prevention (22, 26, 29, 30, 37, 44, 46, 54, 57, 61, 63). “Prevention” is defined as an action that lowers the chance of getting cancer. Ten studies (22, 26, 29, 30, 37, 44, 46, 57, 61, 63) evaluated smoking cessation, and 1 study (54) evaluated uptake of HPV vaccine. Actions to prevent cancer improved following expansion in 8 studies. All of these studies evaluated the prevention of cancers related to tobacco use. These studies revealed increased use of smoking cessation medications. Koma et al. (44) reported a 2.1% increase the probability of smoking cessation in expansion states compared with nonexpansion states. Maclean (46) found that expansion increased Medicaid-financed smoking cessation prescriptions by 34% and led to a 24% increase in new medication use. One study reported that expansion was not associated with increases in smoking quit attempts and smoking cessation after adjusting for state socioeconomic trends, welfare policies, and tobacco control policies (30).

Roberts et al. (54) evaluated trends in HPV vaccination associated with Medicaid expansion. This study concluded that Medicaid expansion as a policy alone was not sufficient to improve HPV vaccine uptake. Expansion in combination with other state policies, including classroom sex education mandates, school-entry requirements, and policies permitting HPV vaccination in pharmacies, could improve vaccine uptake.

Screening

Eighteen studies evaluated access to cancer screening. Most studies evaluated breast cancer screening (n = 12) (18, 19, 22, 33, 37, 40, 43, 45, 53, 56, 57, 62). This was followed by pap smears for cervical cancer (n = 11) (19, 22, 24, 35, 37, 40, 41, 43, 45, 53, 57) and colorectal cancer screening (n = 10) (24, 33, 34, 37, 40, 41, 43, 45, 53, 64). Prostate cancer screening was evaluated in 1 study (55) . Overall, screening increased in 8 studies, no change was reported in 4 studies, and mixed results were found in 6 studies depending on the screening method.

Most studies concluded that Medicaid expansion was not associated with increased access to mammograms (19, 22, 33, 40, 43, 56, 57). Lyu et al. (45) reported statistically significant increases in mammograms in expansion states but only in states with a high supply of primary care providers. Between 2012 and 2013, breast cancer screening among low-income women increased by 4.9% and 3.7% in early-expansion states and nonexpansion states, respectively, but this change was not statistically significant (33).

Seven studies concluded that more women received cervical cancer screening after 2014 (24, 37, 40, 41, 45, 53, 57). Huguet et al. (41) found that cervical cancer screening improved in both expansion and nonexpansion states, with the greatest increase among uninsured patients in expansion states and privately insured patients in nonexpansion states. Four studies reported decreased access to pap smears (19, 22, 35, 43). Alharbi et al. reported that the probability of receiving a pap test decreased statistically significantly among low-income women following Medicaid expansion (19). Two studies concluded there was no change in cervical cancer screening (22, 43).

Eight studies concluded that colorectal screening increased following expansion (33, 34, 37, 40, 41, 45, 53, 64). Two studies found no statistically significant difference in colorectal screening rates (24, 43). In Kentucky, a state that expanded Medicaid in January 2014, screening increased by 27.7% following expansion. Medicaid patients experienced the greatest improvement in screening (292.5%) (34). Fedewa et al. (33) grouped patients into categories based on timing of state’s expansion status: very early (6 states expanding March 2010 to April 2011), early (21 states expanding January 2014 to August 2014), late (5 states expanding January 2015 to July 2016), and not expanding (as of January 2017). Between 2012 and 2016, the proportion of low-income adults up to date with colorectal screening increased by 8.0% in very early states, 2.9% in early states, and 3.8% in nonexpansion states. Among low-income adults in late expansion states, there was no statistically significant change in screening.

Sammon et al. (55) found that although the prevalence of PSA screening decreased overall, low-income residents of early-expansion states experienced a 3% increase in screening. Importantly, the US Preventive Services Task Force supports individualized decision-making for men aged 55-69 years and recommends against screening men 70 years and older (88).

Access to Insurance at the Time of Diagnosis

The most commonly explored outcome was access to health insurance coverage. The majority of studies used a difference-in-differences analysis. Using this methodology, the percentage of uninsured patients may have decreased in both expansion and nonexpansion states, but an overall greater decrease in expansion states (65). Twenty-eight studies concluded that expansion resulted in increased access to insurance (17, 19–25, 27, 28, 36, 38, 39, 41, 42, 44, 47–53, 56–58, 60, 63). In expansion states, 17 studies concluded that Medicaid enrollment increased after passage of the ACA (17–21, 23–25, 27, 36, 41, 42, 49, 50, 52, 58, 60). In 7 studies, fewer patients were enrolled in private insurance (17, 20, 23, 25, 27, 41, 42), whereas 4 studies revealed increases in private-insurance enrollment (21, 24, 56, 58). In nonexpansion states, 4 studies concluded that Medicaid enrollment decreased following implementation of the ACA (17, 23, 42, 60). Access to private insurance increased in 8 studies in nonexpansion states (23, 25, 41, 42, 49, 50, 58, 65).

Before Medicaid expansion, 12.4% of cancer survivors were uninsured. The uninsured rate decreased to 7.7% following ACA implementation (28). Although most of the studies concluded that the percentage of uninsured patients decreased in all states, decreases were greater in expansion than nonexpansion states. Han et al. (38) showed that in the pre-post period, the percentage of uninsured patients declined from 5.2% to 2.6% in expansion states compared with 8.7% to 7.8% in nonexpansion states. Insurance gains in nonexpansion states were related to increased access to private insurance, likely related to other provisions of the ACA (17, 23, 41, 42, 49, 50). Unfortunately, patients with chronic illnesses, including cancer, continue to experience coverage disruptions and loss even in expansion states (36).

Stage at Diagnosis

Eleven studies evaluated whether Medicaid expansion affected the stage at diagnosis (18, 20, 34, 38, 42, 48–50, 56, 58, 59). In the largest study evaluating stage shift, Han et al. (38) reported that expansion was associated with a small but statistically significant increase in the percentage of all early-stage diagnosis for all cancers. Early-stage diagnoses were detected for colorectal, lung, and female breast cancer, and melanoma in both expansion and nonexpansion states. Non-Hodgkin lymphoma and pancreatic cancer were more likely to be diagnosed at an early stage in expansion states, whereas liver cancer was more likely to be diagnosed later (38). Soni et al. (59) found that early-stage diagnoses in Medicaid expansion states were largely driven by an increase in early diagnosis among those aged 35 to 54 years and by cancers amenable to screening (59). In studies evaluating breast cancer in a single expansion state, Ajkay et al. (18) found an increase in early-stage disease in Kentucky, whereas Satyananda et al. (56) reported no difference in diagnosis stage in California. In gynecological malignancies, there was a trend towards more advanced stage disease in uterine and cervical cancer in expansion states. Ovarian cancer staging was not affected by Medicaid expansion (49).

Cancer Incidence

Cancer incidence was evaluated in 5 studies (31, 34, 50, 56, 59). Soni et al. (59) reported that Medicaid expansion was associated with an increase in cancer incidence of 3.4%. The authors conclude that this was driven by an increase in early-stage diagnoses (59). In a similar study evaluating the SEER database, Moss et al. (50) reported an increase in breast and colorectal diagnoses but not lung cancer. In Kentucky, the incidence of colorectal cancer increased after Medicaid expansion in individuals with Medicaid coverage (34). Eberth et al. (31) reported that the mortality-to-incidence ratio for all cancer types was lower in Medicaid expansion states compared with nonexpansion states.

Treatment

Five studies evaluated access to cancer-related treatment, and 1 study evaluated treatment access as a survivor (18, 25, 27, 32, 48, 58). In a study evaluating 81 000 patients who underwent cancer-directed surgery, Crocker et al. (27) found that Medicaid expansion was associated with increased rates of cancer-directed surgery for patients in low-income ZIP codes. There was no change in use in nonexpansion states (27). Similarly, Mesquita-Neto et al. (48) reported that adoption of expansion did not improve access to cancer-directed surgery in the overall cohort but did show improvement in a low-income population. Among people living with HIV and cancer, there was no statistically significant difference in cancer treatment noted between Medicaid expansion and nonexpansion states (25).

Two studies evaluated time to treatment (18, 58). Among women diagnosed with breast cancer in Kentucky, time from diagnosis to surgery increased (P < .01), time from surgery to chemotherapy remained unchanged (P = .26), and time from surgery to radiation decreased (P < .01). Sineshaw et al. (58) reported that time to treatment did not differ between expansion and nonexpansion states among a large cohort of patients with head and neck squamous cell carcinoma. However, among patients with nonoropharyngeal head and neck squamous cell carcinoma, there was a statistically significant improvement in time to treatment (58).

Survivorship

Six studies evaluated the impact of expansion on cancer survivors (28, 31, 34, 36, 39, 51). Among cancer survivors, the probability of having a primary physician increased in both nonexpansion and expansion states by 1.6% and 5.0%, respectively, although the relative change was not statistically significant (51).

Survival Outcomes

Survival outcomes were explored in 4 studies (20, 21, 31, 34). Gan et al. (34) reported that expansion in Kentucky statistically significantly improved colorectal cancer survival in Medicaid enrollees compared with patients pre-ACA implementation, particularly among the Appalachian population. Furthermore, the remaining uninsured patients who did not receive coverage had overall worse survival compared with pre-ACA (34). Barnes et al. (20) reported no change in survival among cancer patients receiving radiation. Two studies reported mixed results (21, 31). Eberth et al. (31) found a lower mortality-to-incidence ratio in Medicaid expansion states in all cancer except those of the esophagus and pancreas. In these more lethal malignancies, the mortality-to-incidence ratio was higher in Medicaid expansion states.

Care Affordability

The impact of Medicaid expansion on the affordability of health-care-related costs was evaluated in 7 studies (37, 39, 43, 44, 46, 51, 53). Six studies reported patient’s perception of affordability and financial barriers to care (37, 43, 44, 51, 53, 66). All studies concluded that there was a decrease in patient-reported concerns related to costs in Medicaid expansion states compared with nonexpansion states.

Analyzing a population-based survey, Han et al. (39) found that adult cancer survivors reported decreased self-reported care unaffordability following implementation of the ACA, with the largest reductions in those residing in expansion states. Nikpay et al. (51) reported that the percentage of survivors who could not afford to see a physician within the last year because of cost fell by 9.1% in expansion states and did not change among nonexpansion states. The percentage of survivors in expansion states who could not afford their medicine also fell. Unpaid medical bills were common in both expansion and nonexpansion states and remained unchanged following changes in Medicaid eligibility. Two studies addressed improved access to smoking cessation and preventive services because of fewer cost-related barriers (44, 46).

Impact on Racial and Socioeconomic Disparities

Twenty-two studies addressed Medicaid expansion’s impact on racial, ethnic, or socioeconomic disparities by comparing changes in access to insurance coverage, screening, and treatment by different demographics (17, 23, 27, 31–33, 36, 38, 39, 42, 43, 45, 47–50, 55, 56, 60–62, 64). Eleven studies concluded that disparities decreased in expansion states, whereas 11 studies found a mixed effect depending on the subgroup analyzed.

Seventeen studies evaluated the impact of expansion on racial or ethnic disparities (17, 23, 27, 32, 33, 36, 38, 39, 47–50, 55, 56, 60, 62, 64). Ten studies found a decrease in disparities (17, 23, 32, 33, 36, 38, 39, 47, 49, 50), 5 studies found a mixed effect (55, 56, 60, 62, 64), and 2 studies concluded no change (27, 48). Han et al. (38) reported that Medicaid expansion resulted in narrowing of disparities among non-Hispanic blacks, Hispanics, and other minority groups. Although there were reductions in the percentage of uninsured patients in nonexpansion states, the overall decrease was small and similar across demographic groups (38). Among surgical cancer patients, Crocker et al. (27) found no statistically significant effect of expansion on racial disparities. Income disparities were evaluated in 18 studies. Fifteen studies concluded that Medicaid expansion was associated with decreased income-related disparities (23, 24, 27, 38, 39, 42, 44, 47–49, 55, 59–61, 64). Three studies found a mixed effect (43, 45, 50). Jemal et al. (42) reported that the percentage of uninsured between low- and high-income groups narrowed substantially in expansion states. This finding is consistent across several other studies evaluating income disparities that concluded that expansion resulted in a narrowing of socioeconomic disparities within expansion states (23, 38, 50).