Estimation of State-Level Prevalence of Hepatitis C Virus Infection, US States and District of Columbia, 2010.

BACKGROUND.: Hepatitis C virus (HCV) infection is the most common chronic blood-borne infection in the United States and a leading cause of morbidity and mortality. Previous analyses of the US National Health and Nutrition Examination Survey (NHANES) indicated approximately 3.6 million noninstitutionalized persons with antibody to HCV (anti-HCV). However, state-level prevalence remains less understood and cannot be estimated reliably from NHANES alone. METHODS.: We used 3 publicly available government data sources to estimate anti-HCV prevalence in each US state among noninstitutionalized persons aged ≥18 years. A small-area estimation model combined indirect standardization of NHANES-based prevalence with logistic regression modeling of mortality data, listing acute or chronic HCV infection as a cause of death, from the National Vital Statistics System during 1999-2012. Model results were combined with US Census population sizes to estimate total number and prevalence of persons with antibody to HCV in 2010. RESULTS.: National anti-HCV prevalence was 1.67% (95% confidence interval [CI], 1.53-1.90), or 3 911 800 (95% CI, 3 589 400- 4 447 500) adults in 2010. State-specific prevalence ranged from 0.71% (Illinois) to 3.34% (Oklahoma). The West census region had the highest region-specific prevalence (2.14% [95% CI, 1.96-2.48]); 10 of 13 states had rates above the national average. The South had the highest number of persons with anti-HCV (n = 1561600 [95% CI, 1 427 700-1 768 900]). The Midwest had the lowest region-specific prevalence (1.14% [95% CI, 1.04%-1.30%]). CONCLUSIONS.: States in the US West and South have been most impacted by hepatitis C. Estimates of HCV infection burden are essential to guide policy and programs to optimally prevent, detect, and cure infection.

Hepatitis C virus (HCV) infection is the most common blood-borne infection in the United States [1]; untreated infection is a leading cause of cirrhosis and hepatocellular carcinoma, is the most frequent cause of liver failure requiring transplantation, and causes more deaths annually than human immunodeficiency virus (HIV) [2-5]. About 70% of persons with untreated HCV infection remain infected for life. HCV infection may be asymptomatic and approximately half of persons are unaware of their infection [6]. Within 30 years, 41% of infected persons progress to cirrhosis, leading to liver failure, hepatocellular carcinoma, and death from liver-related causes [2].

The annual number of new HCV infections in the United States was highest before HCV was discovered to be the cause of “non-A non-B hepatitis” in 1989; incidence declined after prevention guidelines and blood donor screening were implemented in the 1990s. After years of level incidence, new HCV infections began to rise nationally in 2010, increasing >2-fold by 2014. Incidence of acute HCV infections has risen among males, persons aged 20–29 years, and American Indians/Alaska Natives [7]. The prevalence of HCV infection can be measured by a positive test for HCV antibody. Chronic HCV infection is associated with substantial morbidity and mortality, is highest among males and those born during 1945–1965, and has been associated with rising HCV-related mortality in recent years [1, 4, 7–9].

Current US surveillance programs provide incomplete estimates of HCV infection prevalence. In 2014, the Centers For Disease Control and Prevention (CDC) National Notifiable Diseases Surveillance System (NNDSS) included passively collected laboratory reports from 40 states of acute HCV infection, and from 34 states reporting chronic HCV infection. US national and state-specific estimates of HCV infection prevalence cannot be ascertained from NNDSS data [7, 10].

National estimates of HCV prevalence have been produced using the US National Health and Nutrition Examination Survey (NHANES), which includes interviews and physical examinations of noninstitutionalized persons aged ≥6 years [11]. NHANES data for 2003–2010 estimated that 3.6 million persons had antibodies to HCV (anti-HCV), indicative of past or current HCV infection, corresponding to a national prevalence of 1.3% [8]. Subsequent systematic reviews estimate an additional 1 million HCV infections from populations excluded and underrepresented in NHANES, primarily homeless or incarcerated persons [12, 13].

State-level estimates of the prevalence of HCV infection are essential for guiding intervention programs, research, and federal assistance funding priorities among US states. In 2016, CDC funded 7 jurisdictions to conduct active case surveillance for HCV infection, and case counts of HCV diagnoses have been produced [7]. However, these states have been conducting active surveillance for few years, and most states lack local resources to provide similar information regarding the number of persons diagnosed with HCV; surveillance data are limited by variability in testing of persons at risk for infection and reporting to public health authorities.

Some jurisdictions have created their own estimates of prevalence using various methods, most commonly by applying NHANES-derived HCV infection prevalence rates to state populations [14]. The assumption that all states have the national NHANES prevalence of HCV infection might produce inaccurate state-specific estimates, because risk of HCV infection likely varies by state. Some researchers have attempted to improve upon this simple approach by standardizing the national NHANES estimate to local demographic profiles or creating models using case surveillance data (Supplementary Table 1). Despite these informative efforts, there is no complete set of state-specific estimates of HCV infection prevalence for all US states that is based on accurate and consistent methods.

Small-area estimation methods are statistical approaches that can be used for determining state-level HCV infection prevalence by allocating national prevalence estimates into state-specific components using data sources indicating state-level markers of HCV infection prevalence [15]. State-level data sources include electronic medical records (EMRs), insurance claims, and laboratory and mortality data. Individual commercially available EMR, claims, and laboratory datasets are inconsistent in their geographic and demographic representation [16]. Alternatively, mortality is systemically recorded for all decedents in the United States and is publicly available through the population-based National Vital Statistics System (NVSS) [4]. NVSS mortality data provide detailed information regarding decedent age, sex, and race/ethnicity, which are covariates of state-specific HCV infection prevalence.

Here, we describe an application of small-area methodology to deconstruct national NHANES-estimated HCV infection prevalence into state-specific components using HCV-related mortality as a state-level data source. Sensitivity analyses consider the uncertainty of the estimates associated with state-specific variation in diagnosing and treating HCV infection and in recording HCV-related mortality. We present state-level estimates of anti-HCV prevalence and compare these to previously published state estimates.

METHODS

We employed a synthetic small-area estimation approach that combined indirect standardization of NHANES data with regression-based estimates of state-level HCV-related mortality. The data sources and approach are described below and in Table 1 and Figure 1.

Table 1.

Data Sources and Purposes for Primary and Sensitivity Analyses

Data source	Years Represented	Purpose	No. of Individuals Represented	No. of Cases	Data Extraction Notes
National Health and Nutrition Examination Survey	1999–2012	National HCV antibody prevalence overall and by strata of sex, race/ethnicity, and birth cohort	36726 with nonmissing HCV antibody test results	662 with positive anti- HCV test result	NHANES 2000, 2002, 2004, 2006, 2008, 2010, 2012 datasets
US Census intercensal data	1999–2012	Population structure for modeling 14-year average HCV-related mortality rate	3125647447 person-years aged ≥18 y	NA	US Vintage 2000, Vintage 2009, Vintage 2014 datasets
US Census 2010	2010	Population structure for final estimates	234564071 persons aged ≥18 y	NA	US Census 2010 dataset
National Vital Statistics System	1999–2012	Hepatitis C–related mortality, for primary analysis	33540118 decedents age aged ≥18 y who resided in the 50 states or Washington, DC	185285 with HCV as underlying or multiple cause of death	ICD-10 codes included acute viral hepatitis C (B17.1), chronic viral hepatitis C (B18.2)
National Vital Statistics System	1999–2012	Cirrhosis-related mortality, for bias analysis	33540118 decedents aged ≥18 y who resided in the 50 states or Washington, DC	427404 with cirrhosis as underlying or multiple cause of death	ICD-10 codes included hepatic fibrosis (K74.0); hepatic sclerosis (K74.1); hepatic fibrosis with hepatic sclerosis (K74.12); or other and unspecified cirrhosis of liver (K74.6)
National Vital Statistics System	1999–2012	Hepatocellular carcinoma-related mortality, for bias analysis	33540118 decedents aged ≥18 y who resided in the 50 states or Washington, DC	197976 with hepatocellular carcinoma as underlying or multiple cause of death	ICD-10 codes included liver cell carcinoma (C22.0); primary malignant neoplasm of liver, type unspecified (C22.8); or malignant neoplasm of liver, not specified as primary or secondary (C22.9)

Abbreviations: HCV, hepatitis C virus; ICD-10, International Classification of Diseases, Tenth Revision; NHANES, National Health and Nutrition Examination Survey.

Figure 1.

Modeling equations. Abbreviations: anti-HCV, hepatitis C virus antibody; NVSS, National Vital Statistics System.

Data Sources

NHANES 1999–2012

Annually, NHANES uses a complex, multistage sampling design to select approximately 5000 persons in 15 counties, representing the noninstitutionalized population of persons aged ≥6 years residing in the 50 states and Washington, District of Columbia [11].

NHANES tests blood specimens for HCV antibody (anti-HCV), with RNA confirmatory testing [8]. Persons who were anti-HCV positive were considered to have past or current HCV infection [8].

Publicly available data from 7 NHANES release cycles were combined—a 14-year span (1999–2012) that included 36726 observations with anti-HCV test results (662 anti-HCV positive), and self-reported sex, race/ethnicity, and age. Race/ethnicity was categorized into Hispanic, non-Hispanic white, non-Hispanic black, and other (including multiracial) [8]. NHANES respondents were classified into birth cohorts—before, during, and after 1945–1965 [17, 18]. Analyses were restricted to respondents aged ≥18 years; only 5 individuals aged <18 years were anti-HCV positive. The primary analysis did not include RNA confirmatory testing because of missingness for 120 (15%) of tested persons in NHANES datasets.

NVSS Multiple Cause of Death Mortality Data (1999–2012)

Mortality Multiple Cause Microdata files (1999–2012) were obtained through the NVSS [19, 20]. Data were individual records for decedents who lived in a US state or Washington, District of Columbia and were aged ≥18 years at death. Records also contained International Classification of Diseases, Tenth Revision (ICD-10) codes for multiple underlying causes of death (N = 33540118). Demographic covariates included sex, race/ethnicity, and birth cohort. In the primary analysis, any records including the ICD-10 code for acute viral hepatitis C (B17.1) or chronic viral hepatitis C (B18.2) as an underlying cause of death were considered to signal HCV-related mortality (n = 185285).

US Census Intercensal Data (1999–2012) and 2010 US Census Data

Intercensal population estimates from the US Vintage 2000, 2009, and 2014 datasets provided denominators for HCV-related mortality rates during 1999–2012 [21]. Data were grouped into the same sex, race/ethnicity, and birth cohort categories as the NHANES and NVSS data. State-by-demographic group population totals from the 2010 US Census were used to calculate 2010 state HCV prevalence [21].

Analysis

The number of persons in each state with anti-HCV was computed using the standardization-based estimator in Figure 1, equation 1. First, we calculated direct weighted estimates of national HCV-antibody prevalence for 24 strata (sex × race/ethnicity × birth cohort), using standard methodology (Figure 1, equation 1) [15]. We multiplied weighted estimates by state-by-demographic stratum 2010 population counts to generate crude state-level estimates. These were adjusted by the ratio of state-by-demographic stratum effects, based on the average HCV-related death rate in the 14-year period. We fit a high-order logistic regression model that approximated full stratification (several of 1224 strata had zero cells), permitting detection of heterogeneity among strata (Figure 1, equation 3). We assessed collinearity and model fit by comparing observed state-level HCV-related mortality totals to model predictions [15, 22]. Mortality-adjusted HCV infection prevalence totals were summarized to yield estimated state-level totals (rounded to nearest hundred persons), with prevalence rates (Figure 1, equations 1 and 2). Supplementary analyses estimated state-level chronic HCV infection, defined as a positive or indeterminate anti-HCV test and a positive HCV RNA test, using the above approach but with a 12-stratum model (race/ethnicity considered white non-Hispanic or not), due to more sparse NHANES data.

Sensitivity Analyses

To account for the joint statistical uncertainty in the stratified NHANES estimates and model-based HCV-related mortality estimates, we conducted a Monte Carlo simulation that respectively sampled from logit-normal and normal distributions (k = 10000 runs), using the standard errors for the original estimates, to produce 95% confidence intervals (CIs) for state-level estimates.

There might be state-level variability in HCV diagnosis and treatment that produces state-level variability in HCV-related deaths or proper attribution (specific ICD-10 codes) of deaths to an HCV cause, although likely limited [5]. We repeated all analyses with a broader definition of HCV-related deaths that used a combination of the HCV-specific ICD-10 codes and less-specific, more sensitive, ICD-10 codes representing cirrhosis-related and hepatocellular carcinoma–related (HCC) causes of death (Table 1) [5]. To increase specificity of these additional codes, we applied available estimates of the population attributable fraction (PAF) due to HCV infection (cirrhosis: 42%, HCC: 48%; Figure 1, equation 4) [23]. We secondarily considered other PAF estimates in less representative populations [24, 25].

We descriptively compared model findings to other publicly available reports of state estimates. Reports were excluded if HCV infection estimates solely involved applying national NHANES HCV infection prevalence estimates to total state population or if estimation was exclusively based on partial case surveillance data for HCV infection. Reports not describing the methodology used for prevalence estimation were included to facilitate more comparisons. Where possible, abstracted prevalence estimates were restricted to comparable noninstitutionalized populations.

RESULTS

The estimated national prevalence of anti-HCV in 2010 was 1.67% (95% CI, 1.53%–1.90%), corresponding to 3 911 800 (95% CI, 3589400–4447500) US adults with past or current HCV infection. Demographic stratum-specific estimates ranged from 0.26% for other race/ethnicity females born after 1965 to 8.02% for black males born 1945–1965 (Supplementary Table 2).

The prevalence of anti-HCV varied by state (Table 2 and Figure 2). State-specific prevalence rates ranged from 0.71% in Illinois to 3.34% in Oklahoma, with >2.5% additionally in District of Columbia, New Mexico, Oregon, and Tennessee. By census region and division, the West contained the most high-prevalence states: 10 of 13 states were above the national average, and the region-specific prevalence was 2.14%. The Pacific West division was most affected (2.30% prevalence; 865400 persons who had been infected with HCV). The South had the second-highest anti-HCV prevalence (1.80%) and the largest number of persons with anti-HCV (n = 1561600). Within the South, the West South Central had the highest prevalence (2.19%), while the South Atlantic division had the most persons with anti-HCV (n = 712900). The Midwest (1.14%) and Northeast (1.43%) had relatively lower anti-HCV prevalence. Of the 21 states in these regions, only Rhode Island had prevalence above the national average (2.12%). Although having prevalence rates lower than the national average, Michigan, New York, Pennsylvania, and Ohio each had >100000 persons with anti-HCV.

Table 2.

Estimated Total and Prevalence Rate of Persons With Hepatitis C Virus Antibody, US States and District of Columbia, by US Census Region and Division, 2010a

Region/Division/State	2010 Census Population	Total Persons With Anti-HCV		Anti-HCV Prevalence Rate
	No.	No.	(95% CI)	Rate per 100	(95% CI)
NORTHEAST	42984048	614300	(565700–693900)	1.43	(1.32–1.62)
New England	11293971	157800	(143400–182100)	1.40	(1.27–1.61)
Connecticut	2757082	36800	(33500–42400)	1.33	(1.21–1.54)
Maine	1053828	11200	(9600–13900)	1.06	(0.91–1.32)
Massachusetts	5128706	74100	(66900–85600)	1.44	(1.30–1.67)
New Hampshire	1029236	11000	(9300–13700)	1.07	(0.91–1.33)
Rhode Island	828611	17500	(15600–21000)	2.12	(1.89–2.53)
Vermont	496508	7200	(6100–9400)	1.45	(1.23–1.89)
Middle Atlantic	31690077	456500	(421300–513200)	1.44	(1.33–1.62)
New Jersey	6726680	90700	(83000–103000)	1.35	(1.23–1.53)
New York	15053173	223700	(207000–252700)	1.49	(1.38–1.68)
Pennsylvania	9910224	142100	(129200–160900)	1.43	(1.30–1.62)
MIDWEST	50798893	578600	(527700–659200)	1.14	(1.04–1.30)
East North Central	35269776	378800	(345200–429200)	1.07	(0.98–1.22)
Indiana	4875504	59100	(52600–68300)	1.21	(1.08–1.40)
Illinois	9701453	68400	(62500–78000)	0.71	(0.64–0.80)
Michigan	7539572	101200	(92000–114600)	1.34	(1.22–1.52)
Ohio	8805753	119000	(107700–135000)	1.35	(1.22–1.53)
Wisconsin	4347494	31100	(28000–36200)	0.71	(0.64–0.83)
West North Central	15529117	199800	(181700–231100)	1.29	(1.17–1.49)
Iowa	2318362	24600	(21700–29500)	1.06	(0.94–1.27)
Kansas	2126179	29900	(26600–35200)	1.41	(1.25–1.66)
Minnesota	4019862	41500	(37600–48500)	1.03	(0.94–1.21)
Missouri	4563491	76900	(68900–88300)	1.69	(1.51–1.93)
Nebraska	1367120	16100	(14300–19500)	1.18	(1.05–1.42)
North Dakota	522720	4400	(3700–6200)	0.83	(0.71–1.19)
South Dakota	611383	6300	(5400–8700)	1.03	(0.88–1.42)
SOUTH	86766987	1561600	(1427700–1768900)	1.80	(1.65–2.04)
South Atlantic	46020646	712900	(652100–805700)	1.55	(1.42–1.76)
Delaware	692169	13600	(12200–16100)	1.97	(1.76–2.33)
District of Columbia	500908	16400	(14400–19500)	3.27	(2.87–3.90)
Florida	14799219	245600	(221700–280700)	1.66	(1.50–1.90)
Georgia	7196101	84500	(76800–96100)	1.17	(1.07–1.34)
Maryland	4420588	82000	(74900–93100)	1.86	(1.69–2.11)
North Carolina	7253848	117300	(106600–133100)	1.62	(1.47–1.83)
South Carolina	3544890	62300	(56200–71400)	1.76	(1.58–2.01)
Virginia	6147347	66700	(60800–76000)	1.09	(0.99–1.24)
West Virginia	1465576	24400	(21000–29500)	1.66	(1.43–2.01)
East South Central	14025119	264300	(236500–303300)	1.88	(1.69–2.17)
Alabama	3647277	52400	(47000–60400)	1.44	(1.29–1.66)
Kentucky	3315996	54200	(47200–63800)	1.63	(1.42–1.92)
Mississippi	2211742	35200	(31300–40900)	1.59	(1.42–1.85)
Tennessee	4850104	122500	(108900–141100)	2.53	(2.25–2.91)
West South Central	26721222	584400	(536000–664500)	2.19	(2.01–2.49)
Arkansas	2204443	37500	(33000–43900)	1.70	(1.50–1.99)
Louisiana	3415357	76200	(69300–86400)	2.23	(2.03–2.53)
Oklahoma	2821685	94200	(83800–112900)	3.34	(2.97–4.00)
Texas	18279737	376600	(345900–428000)	2.06	(1.89–2.34)
WEST	54014143	1157400	(1060100–1341100)	2.14	(1.96–2.48)
Mountain	16368084	291900	(266100–339700)	1.78	(1.63–2.07)
Arizona	4763003	90000	(81400–104600)	1.89	(1.71–2.20)
Colorado	3803587	66100	(60000–76500)	1.74	(1.58–2.01)
Idaho	1138510	16400	(14200–20200)	1.44	(1.25–1.77)
New Mexico	1540507	42600	(37900–51400)	2.76	(2.46–3.34)
Montana	765852	14900	(12900–19300)	1.94	(1.69–2.52)
Utah	1892858	17600	(15500–21300)	0.93	(0.82–1.13)
Nevada	2035543	36500	(32800–42700)	1.80	(1.61–2.10)
Wyoming	428224	7800	(6700–10200)	1.83	(1.56–2.39)
Pacific	37646059	865400	(792300–1002900)	2.30	(2.10–2.66)
Alaska	522853	11400	(9700–16400)	2.19	(1.85–3.13)
California	27958916	629600	(578800–726100)	2.25	(2.07–2.60)
Hawaii	1056483	15700	(13600–20300)	1.48	(1.29–1.92)
Oregon	2964621	90500	(79800–106800)	3.05	(2.69–3.60)
Washington	5143186	118300	(105800–138800)	2.30	(2.06–2.70)
50 US States & Washington, DC	234564071	3911800	(3 589 400–4 447 500)	1.67	(1.53–1.90)

Abbreviations: CI, confidence interval; Anti-HCV, hepatitis C virus antibody.

aDefined as persons with HCV antibodies, indicating past or current hepatitis C infection.

Figure 2.

Estimated total persons with hepatitis C virus antibody (anti-HCV; indicating past or current HCV infection) (A) and anti-HCV prevalence rates (B), United States and District of Columbia, 2010.

In sensitivity analyses that considered the addition of 2 potential indicators of HCV-related deaths, the above patterns were preserved, generally yielding state-level estimates closer to the national average (Figure 3 and Supplementary Tables 3–4). Inclusion of cirrhosis-related mortality resulted in 39 of 51 state estimates within the 95% CI bounds for the primary estimates that used only HCV-specific mortality. The median absolute deviation of state estimates between these 2 estimation approaches was 8% (interquartile range [IQR], 5%–16%, multiplicative scale). Inclusion of HCC-related mortality resulted in 50 of 51 state estimates that were within the 95% CI bounds for the primary estimates using HCV-specific mortality. The median absolute deviation of state estimates between these 2 estimation approaches was 3% (IQR, 2%–6%). Results were similar when considering alternative, but less representative PAF estimates (Supplementary Tables 5–6). There was high agreement between the primary anti-HCV prevalence estimates and 11 external estimates of state-level HCV infection prevalence (Supplementary Table 1). The 2 external estimates that utilized state-level HCV testing data (Arkansas, Oregon) were closest to our primary estimates (5.5% and 4.8% higher, respectively). To facilitate comparisons, Supplementary Tables 7–8 display model-based chronic HCV infection and NHANES standardization–based anti-HCV estimates.

Figure 3.

Estimated hepatitis C virus antibody (anti-HCV) prevalence rates from sensitivity analyses using additional HCV-related mortality International Classification of Diseases, Tenth Revision codes. Abbreviations: anti-HCV, hepatitis C virus antibody; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; PAF, population attributable fraction.

DISCUSSION

We used a small-area estimation approach to synthesize data from 3 publicly available systematic, population-based data systems and provide the first comprehensive state estimates of the prevalence of anti-HCV in the United States. The estimates were robust to multiple approaches based on additional mortality data, and were comparable to independent state estimates. The method and its results can be applied by state and local health officials to guide program planning, set priorities for resource allocation, and evaluate interventions.

These state-specific estimates bring a new level of understanding to our prior knowledge about the epidemiology of HCV infection in the United States. States with high estimated anti-HCV prevalence also contain highly urban populations (eg, Rhode Island, District of Columbia) and high proportions of the population who inject drugs (eg, Vermont, Tennessee, West Virginia) or who are Native Americans (eg, Oklahoma, New Mexico; the model did not explicitly include Native American race) [26, 27]. The current anti-HCV prevalence estimates highlight states which, by their demographic structure alone, might be predicted to have lower prevalence. For example, simple NHANES standardization yields 44000 estimated cases in Oklahoma; incorporating additional data on HCV-related deaths in Oklahoma suggests an estimated prevalence more than twice that high.

Our approach has significant methodologic strengths, as the data sources are population-based and representative of the underlying populations. However, our analysis has important limitations. We aggregated data across multiple years of NHANES to achieve a sufficient sample for stable estimates. It is possible that there were secular trends over the NHANES period. The NHANES population samples noninstitutionalized adults and therefore excludes incarcerated persons, homeless persons, those in active military service, and persons on tribal lands. These populations are critical constituents of the US HCV epidemic, comprising up to a fifth of prevalent antibody- positive persons in the United States [12]. Thus, our estimates apply only to noninstitutionalized US populations. Mortality data used to allocate the NHANES-based anti-HCV total included deaths from institutionalized populations. Complete laboratory data in NHANES were nearly universally available for the anti-HCV test, but were less complete for RNA testing. Thus, our primary estimates of persons with anti-HCV includes the 15%–25% of people infected by HCV who may have cleared their HCV infections but retained detectable antibodies [28]. Antibody-based prevalence of individuals previously or currently infected with HCV provides a conservative overestimate of persons indicated for clinical evaluation and HCV-related medical services [8]. The HCV RNA test more accurately indicates active infection needing treatment. Importantly, the proportion of anti-HCV positive persons with detectable HCV by polymerase chain reaction is also influenced by the number of persons successfully treated and cured of their HCV infection. Over time, this proportion should decline as more persons are identified, treated, and cured; however, the proportion will vary among populations with differential access to HCV-related care and treatment.

Finally, NHANES is likely less representative of populations of persons infected in recent years. Incidence of HCV infection has been increasing since 2010, and those infected in more recent epidemics are younger, more likely to live in rural areas, and likely to acquire HCV through needle-sharing behaviors associated with opioid use [29]. Nationally, the number of recent new infections is small relative to the prevalent population, although some states with modest or low prevalence of infection have had larger increases than reflected in national trends [7]. Methods that incorporate recent trends in incident HCV infection and in treatment can improve future iterations of this model.

There is some mismatch between the NHANES and mortality data: mortality data represent older trends in the epidemic. Also, HCV infection may not be consistently diagnosed or recorded on death certificates by state. However, this low sensitivity of identified HCV infection in death records should not cause bias in our estimates unless the likelihood of diagnosis or recording on death certificates varies across states. We posit that deficiencies in diagnosis or recording likely occur, but are at a facility level and do not vary systematically by state. The impact of varying sensitivities of our case definition in mortality data were explored in extensive sensitivity analyses, and did not result in substantial changes to our estimates.

CONCLUSIONS

Although national recommendations for HCV prevention, testing, and clinical management are developed by CDC and other authorities [30], decisions regarding the capacity to deliver these services are made at the state level. State-level estimates can inform these decisions in multiple ways. First, these data can prompt reconsiderations of HCV disease burden. These modeled prevalence estimates are based on publicly available data, allowing local authorities to assess biases in data sources used in the model (ie, mortality data) and the representativeness of the modeled data in comparisons with other local data sources (ie, HCV surveillance). Second, these results, supported by local public health authorities, provide new information to engage stakeholders, resulting in agreed-upon state/local estimates of HCV infection prevalence. State-level prevalence estimates allow state and local health officials to consider more investments in surveillance and collection of other strategic data for refining our estimates. These data can also be used to revise HCV-related prevention plans and guide prevention initiatives. For example, these data can help state/local health officials estimate the number of HCV-infected persons who remain undiagnosed. The results may help state Medicaid programs to budget funds for HCV testing and treatment. More broadly, having state-level estimates calculated consistently across states will allow states to assess their standing in relation to other states and to the nation as a whole, and to adapt their prevention and control efforts to national or other state programs that have been shown to be effective.

The approach reported here can be routinely updated as new data become available. The same approach to estimation can also be extended to yield estimates for demographic strata, areas smaller than states, and excluded populations in each state. Modeling approaches can serve as a useful method to quantify the HCV epidemic in the absence of national case surveillance data. However, this should not minimize the imperative of continuing to enhance local surveillance efforts, especially in the context of curative HCV therapies. Surveillance data may provide more accurate and reliable estimates of the burden of HCV infection and serve as the basis for public health programs to diagnose persons living with hepatitis C and link them to appropriate clinical services and treatment.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Click here for additional data file.