Drug and Alcohol Dependence 153 (2015) 86–93

Contents lists available at ScienceDirect

Drug and Alcohol Dependence journal homepage: www.elsevier.com/locate/drugalcdep

Health insurance coverage and healthcare utilization among infants of mothers in the national methadone maintenance treatment program in Taiwan Shao-You Fang a , Nicole Huang a , Ting Lin a , Ing-Kang Ho b,c , Chuan-Yu Chen a,d,e,∗ a

Institute of Public Health, National Yang-Ming University, Taiwan Center for Drug Abuse and Addiction, China Medical University, Taiwan c Graduate Institute of Clinical Medicine, China Medical University, Taiwan d Center of Neuropsychiatric Research, National Health Research Institutes, Taiwan e College of Public Health, National Taiwan University, Taiwan b

a r t i c l e

i n f o

Article history: Received 5 August 2014 Received in revised form 30 May 2015 Accepted 31 May 2015 Available online 10 June 2015 Keywords: Access to health care Children Addiction

a b s t r a c t Background: Children of heroin-using women have a higher risk of unfavorable health and developmental outcomes. Although methadone maintenance treatment (MMT) has been widely used to treat heroinusing pregnant women, potential effects on accessibility and utilization of healthcare service for their offspring are less explored. Methods: We used four national registry and health insurance datasets in Taiwan from 2004 to 2009 to form a population-based matched retrospective cohort study. A total of 1056 neonates born to women in the MMT program (857 born before mother’s enrollment in the MMT program [BM], 199 born after mother’s enrollment in the MMT program [AM]) was established; 10 547 matched non-drug [ND] exposed neonates were identified for comparison. Outcome variables included offspring’s health insurance coverage and utilization of preventive, outpatient, and emergency room cares in the first year after birth. Results: Infants born to mothers on MMT were more likely to have no or incomplete insurance coverage (BM: adjusted odds ratio [aOR] = 1.29, 95% CI: 1.10–1.53; AM: aOR = 1.56, 95% CI: 1.14–2.13) as compared with the socioeconomic status-matched ND group. The BM infants appeared to have fewer preventive care visits (adjusted relative risk [aRR] = 0.85, 95% CI: 0.80–0.90), whereas the AM infants utilized outpatient and emergency room services more frequently (outpatient: aRR = 1.11, 95% CI: 1.01–1.23; emergency: aRR = 1.46, 95% CI: 1.11–1.90). Conclusions: Addiction treatment and harm reduction programs for women of childbearing ages should be delivered in the coordinated framework that ensures comprehensiveness and continuity in healthcare and social services. © 2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Substance use during pregnancy is an important health and social problem. To date, it is estimated that less than 1% of pregnant women aged 15–44 in the United States and nearly 6% in South Australia used illegal drugs during pregnancy, with heroin being one of the most commonly reported drugs (Kennare et al., 2005; Substance Abuse and Mental Health Services Administration,

∗ Corresponding author at: Institute of Public Health, Medical Building II, Rm 204, National Yang-Ming University, No. 155, Sec. 2, Linong Street, Taipei 112, Taiwan. Tel.: +886 2826 7000x7344; fax: +886 037 583 461. E-mail address: [email protected] (C.-Y. Chen). http://dx.doi.org/10.1016/j.drugalcdep.2015.05.044 0376-8716/© 2015 Elsevier Ireland Ltd. All rights reserved.

2012). Adverse effects of opioid use and abuse during pregnancy can be manifested in pregnancy process (e.g., placental abruption and miscarriage; Hulse et al., 1998a; Kennare et al., 2005), or in trans-generational unfavorable perinatal outcomes on offspring (e.g., preterm birth, low birth weight, and neonatal death; Burns et al., 2006; Hulse et al., 1998b; Kennare et al., 2005). In many countries, methadone maintenance treatment (MMT) is the first line of management to help pregnant women with heroin addiction ameliorating intoxication and withdrawal. Cumulative evidence suggests that methadone is beneficial to heroin-dependent pregnant women in terms of pregnancy outcomes; however, there is still a great concern about the adverse effects of methadone on neonates’ health outcomes (Hulse et al., 1998a, b; Kennare et al., 2005; Webster et al., 1996). Some clinical

S.-Y. Fang et al. / Drug and Alcohol Dependence 153 (2015) 86–93

observational studies have found that certain unfavorable neonatal outcomes, such as growth retardation, low birth weight, preterm delivery, and neonatal abstinence syndrome [NAS], were very common (e.g., NAS: 40–75%) among the offspring of women enrolled in the opioid substitute therapy (Burns et al., 2010; Chen et al., 2015; Cleary et al., 2012; Dryden et al., 2009; Hulse et al., 1997, 1998b; Kakko et al., 2008). Relative to infants of non-drug using mothers, the needs in healthcare for children born to women on an opioid substitute therapy are expected to be higher (Johnson et al., 2003; Jones et al., 2010; Kakko et al., 2008), and as such their access to quality and regular well child services is especially important. Health insurance coverage and healthcare utilization have been conceptualized as two important dimensions when evaluating children’s access to health care. Insurance coverage is often identified as the most crucial determinant for health care utilization, especially in countries without universal coverage (Abdus and Selden, 2013; Cummings et al., 2009; Dietz et al., 2012; Holl et al., 1995; Kogan et al., 2010). Evidence from the United States consistently indicates that none- or discontinuously insured children, as compared with fully insured ones, were more likely to experience delays in receiving needed health services and to have difficulties in accessing a usual source of care, including preventive healthcare (Cummings et al., 2009; Holl et al., 1995; Olson et al., 2005). Given the importance of health insurance coverage, some strategies have been implemented at national, state, or local levels to reduce the access barriers (e.g., universal health care, social insurance, and special program), yet the related benefits can still be underutilized. For example, studies on children eligible for the public health insurance program in the US have shown that over one in ten children were still uninsured (DeVoe et al., 2008, 2011). Factors influencing children’s insurance status and health service use included parents’ insurance status, employment, educational attainment, copayment, rural residence, among others (DeVoe et al., 2008, 2011; Quimbo et al., 2008). The first year of life is the period with the greatest need for healthcare. Evidence indicates that having up-to-date pediatric preventive care (e.g., well-child visits) was associated with reduced avoidable hospitalization among poor and near-poor children (Hakim and Bye, 2001), and that having better access to primary health care may lower children’s visits to emergency department (Piehl et al., 2000). Therefore, to monitor growth/development and to provide timely treatment, pediatric preventive and primary health care visits are crucial (Magnani et al., 1996), and this is especially true for those infants in poor health and disadvantaged families. Prior studies have shed light on health care utilization in the first year of life in a variety of subpopulations, such as children of low income families and children with special needs (Dietz et al., 2012, 2013; Farr et al., 2013; Holl et al., 2012). However, information is generally scant on offspring of heroin-addicted or methadone-treated mothers even though this group of children often experienced more health and developmental problems due to in utero heroin or substitute medication exposure, maternal unfavorable lifestyle, and disadvantaged socioeconomic condition. In Taiwan, the National Health Insurance Program (NHIP) has been implemented since 1995 and has provided comprehensive and universal health insurance coverage for more than 23 million enrollees. Although the NHIP is compulsory, the coverage rate of NHIP has not reached 100% due to job loss or change or incarceration for more than two months. Since children under 18 are generally covered as dependents under their employed parents, parental unemployment and financial disadvantage may affect their insurance status and healthcare utilization. To address the abovementioned research gaps, we conducted a retrospective longitudinal study using several national datasets in Taiwan to investigate the accessibility and utilization of healthcare among children born to heroin-addicted women. Specifically, we examined (i)

87

whether maternal methadone treatment is an important factor when assessing children’s insurance coverage in the first year of life under the universal health insurance program, and (ii) whether the utilization of preventive, outpatient, and emergency care in the first year of life is affected by maternal methadone treatment status. 2. Methods 2.1. Multiple sources of datasets The administrative data used in this study came from (i) the 2004–2008 Birth Notification System in the Bureau of Health Promotion, (ii) the 2004–2008 Birth Registry in the Ministry of the Interior, (iii) the 2006–2009 MMT program in the Center for Disease Control and Prevention, (iv) the 2001–2009 Death Registry in the Ministry of Health and Welfare, and (v) the 2001–2009 National Health Insurance Database in the National Health Insurance Administration. The dataset was linked with each other via encrypted maternal or children’s identification number (ID).

2.1.1. The Birth Notification System and the Birth Registry. In Taiwan, the Birth Notification System contains information on both live births and stillbirths, whereas the Birth Registry records only live births. Both live births and stillbirths of neonates weighing at least 500 g or beyond 20 weeks of gestation have to be reported to the Bureau of Health Promotion through the Birth Notification System within seven days of the birth. The information is recorded under the mother’s name, and is managed by the Bureau of Health Promotion. For live births, the parents or relatives can register the birth with the local household registry office using the birth certificate issued by the hospital within 60 days of the birth. The local household registry office is required to forward the registration information to the central Birth Registry system. Since the ID of a child is unavailable in the Birth Notification System, we linked the Birth Notification System with the Birth Registry via encrypted maternal ID to retrieve the children’s ID for subsequent analyses. Through this procedure, an estimated 88% of the 1 039 569 live births in the Birth Notification System were linked with the Birth Registry during the years 2004–2008. Important children’s and maternal characteristics were retrieved from the Birth Notification System (e.g., preterm delivery and age at delivery) and the Birth Registry (e.g., maternal marital status; Chen et al., 2015).

2.1.2. The MMT dataset. To prevent the spread of HIV infection through needle sharing among heroin addicts, the Taiwanese government launched the AIDS harm reduction program in 2005 and the program has been implemented nationwide since August, 2006 by the Center for Disease Control and Prevention. In this substitute therapy program, heroin addicts were given oral methadone daily in the MMT clinics (take-home methadone is not allowed). The MMT dataset includes information pertaining to methadone treatment during 2006–2009 (e.g., entry date of treatment). Via data linkage of the Birth Notification System, the Birth Registry, and the MMT, we identified neonates born before and after maternal participation in MMT. The linkage rates between the Birth Notification System and the Birth Registry were 96.51% for the MMT group and 88.6% for the non-MMT group.

2.1.3. The National Health Insurance Database. The National Health Insurance Database was initiated by the NHIP and managed by the National Health Insurance Administration. Information concerning insurance status and health care utilization among mothers and children was retrieved from the 2001 to 2009 National Health Insurance Database.

2.2. Study population In this study on accessibility and utilization of healthcare during the first year of life, we defined children who were born between 2004 and 2008 and have lived to the first birthday as the study population. According to the initial enrollment date in the MMT and the date of delivery, infants born to mothers in the MMT were subdivided into two groups: those born before the enrollment (BM: n = 857; the average interval between delivery and MMT enrollment is 2.4 years) and after (AM: n = 199; the third quartile compliance rate through pregnancy was estimated 60%) (see Fig. 1 for data linkage process; Chen et al., 2015). In this study, the children in the BM group are assumed to be heroin-exposed since empirical evidence indicated that average year of heroin history prior to MMT was 5 years in Taiwan (Lin et al., 2013). For comparison, non-drug exposed [ND] infants were defined as neonates born to mothers who had never had substance use problems (International Classification of Diseases, 9th Version, Clinical Modification (ICD-9-CM) code: 291–293 and 304–305) within two years before their delivery. To increase analytic efficiency, a 1:10 ratio of matching was used on the criteria of birth year and month (2004–2008), maternal age at delivery, and maternal insurance premium at delivery (n = 10 547 ND infants). The Institutional Review Board (IRB) of the National Health Research Institutes approved this research (IRB No. EC0990603-E).

88

S.-Y. Fang et al. / Drug and Alcohol Dependence 153 (2015) 86–93

Fig. 1. Study population and data linking process.

2.3. Definition of primary independent and outcome variables The primary independent variable in this study is the enrollment in the MMT, which was classified into BM (born before maternal methadone treatment), AM (born after maternal methadone treatment), and ND (non-drug exposed) groups. Outcomes in this study included insurance coverage and health care utilization within one year of the birth. The insurance coverage was classified into 0–11 months (discontinuous or no coverage) and 12 months (complete coverage). Health care utilization was measured by the number of preventive care, outpatient, and emergency room (ER) visits before the child’s first birthday. 2.4. Measured covariates Other important covariates included maternal (e.g., educational attainment) and infants’ characteristics (i.e., gender and preterm delivery). Maternal age at delivery was divided into four groups: 25 and below, 26–30, 31–35, and over 35 years old. Since the insurance premium is calculated based on income, it serves as a proxy for maternal social economic status at delivery and was subdivided into five categories: poverty (i.e., low income), near poverty (i.e., monthly income below 17 000 New Taiwanese Dollar [NTD]), medium premium (i.e., NTD 17 001–35 000), high premium (more than NTD 35 000), and non-insured (Chen et al., 2010). Educational attainment was classified into elementary school, middle/high school, and college or above. Maternal marital status was determined by the information about the infant’s father in the Birth Registry. Urbanicity of maternal residence was defined on the basis of the urbanization index for Taiwan’s city/county (Liu et al., 2006), and regrouped into urban, suburban, and rural areas. Number of prior deliveries was assessed from the maternal outpatient and inpatient records for delivery within three years before the index child was born (ICD-9-CM code: 650–672). The NHIP reimburses patients for up to ten prenatal care visits, scheduled according to the gestational age. We calculated the number of reimbursed prenatal visits and grouped them as follows: 0, 1–3, 4–6, and ≥7. Preterm delivery was defined as less than 37 weeks of gestational age. As to maternal mental disorders during postpartum period, we coded mental disorders (ICD-9-CM code: 290, 294–303, 306–319) as positive when mothers visited outpatient care for these disorders at least once within one year of the delivery. 2.5. Statistical analyses Frequencies and proportions were first used to understand the distribution of characteristics among BM-, AM-, and ND-infants, and chi-square tests were performed to compare between the MMT (BM or AM) and ND groups. Logistic regression

models were employed to examine the relationship between MMT participation and insurance coverage. Since children without insurance coverage before their first birthday did not have any insurance-reimbursed health care in the NHIRD, we decided to drop those without full insurance coverage before the first birthday in the subsequent analyses on health care utilization (n = 752). After examining the number of health care visits across three groups (mean and variance), we decided to use the negative binomial regression model to accurately estimate the association between MMT participation and preventive and outpatient care utilization. As to emergency care visits, we used zero-inflated negative binomial regression analysis to assess the association since approximately 60–70% of infants in each group did not have any ER visit in their first year of life. All the data linkage and statistical analyses were performed by researchers at the Collaboration Center of Health Information Application (CCHIA) at the Ministry of Health and Welfare. Any number of observations less than five was not included. All analyses were carried out using SAS 9.3 (SAS Institute Inc., Cary, NC).

3. Results The characteristics of the BM (born before the MMT enrollment)-, AM (born after the MMT enrollment)-, and ND (nondrug exposed)-infants are summarized in Table 1. As compared to the ND-group, a relatively higher proportion of infants in the BM-group were born to mothers who had lowest or highest levels of educational attainment (i.e., elementary school: 12.02% vs 3.74%; college or above: 61.26% vs 48.98%, P < 0.001), unmarried status (60.21% vs 19.92%, P < 0.001), mental disorders in postnatal period (15.64% vs 5.20%, P < 0.001), and fewer prenatal visits (0 visit: 25.90% vs 3.80%, P < 0.001). As for infants’ characteristics, preterm deliveries (11.55% vs 4.53%, P < 0.001) and incomplete insurance coverage (0–11 months: 33.84% vs 25.17%, P < 0.001) appeared more common in those in the BM-group. A similar pattern also manifested in the comparison between the AM-group and ND-group (see middle and right panels in Table 1). The mother- and child-associated factors for no/incomplete insurance coverage in the child’s first year of life are presented in Table 2. Compared to the income-matched ND infants, those born

S.-Y. Fang et al. / Drug and Alcohol Dependence 153 (2015) 86–93 Table 1 Characteristics of infants born between 2004 and 2008, by maternal methadone treatment history, 2004–2009, Taiwan.

Maternal characteristics Age at delivery (years) ≤25 26–30 31–35 >35 Maternal insured amount at deliverya Higher amount Middle amount Near poverty Poverty Non-insured Educational attainment Elementary Mid/high school College or above Marital status Married Unmarried Urbanization of residence Urban Suburban Rural Number of prior deliveryb 0 1 ≥2 Number of prenatal examination 0 1–3 4–6 ≥7 Outpatient visits for mental disorders after deliveryc Yes No Child’s characteristics Gender Boy Girl Preterm delivery (≤36/weeks) Yes No Insurance period within one year after birth 12 months 1–11 months 0 months

Before methadone (n = 857), no. (%)

After methadone (n = 199), no. (%)

Non-drug exposed (n = 10 547), no. (%)

245 (28.58) 362 (42.24) 176 (20.54) 74 (8.63)

32 (16.08) 80 (40.20) 67 (33.67) 20 (10.05)

2764 (26.20) 4419 (41.90) 2424 (22.98) 940 (8.91)

9 (1.05) 199 (23.22) 636 (74.21) 6 (0.70) 7 (0.82) *** 103 (12.02) 229 (26.72) 525 (61.26) *** 341 (39.79) 516 (60.21) **

0 (0.00) 39 (19.60) 160 (80.40) 0 (0.00) 0 (0.00) *** 22 (11.06) 56 (28.14) 121 (60.80) *** 75 (37.69) 124 (62.31)

110 (1.04) 2360 (22.38) 7920 (75.09) 70 (0.66) 87 (0.82)

138 (16.10) 595 (69.43) 48 (5.60) **

31 (15.58) 145 (72.86) 11 (5.53) ***

2341 (22.20) 7384 (70.01) 668 (6.33)

686 (80.05) 161 (18.79) 10 (1.17) ***

177 (88.94) 22 (11.06) 0 (0.00) ***

7978 (75.64) 2358 (22.36) 211 (2.00)

222 (25.90) 193 (22.52) 218 (25.44) 224 (26.14) ***

54 (27.14) 49 (24.62) 35 (17.59) 61 (30.65) ***

401 (3.80) 701 (6.65) 2375 (22.52) 7070 (67.03)

134 (15.64) 723 (84.36)

36 (18.09) 163 (81.91)

548 (5.20) 9999 (94.80)

393 (3.74) 4972 (47.28) 5150 (48.98) 8446 (80.08) 2101 (19.92)

483 (56.36) 374 (43.64) ***

108 (54.27) 91 (45.73) ***

5613 (53.22) 4934 (46.78)

99 (11.55) 758 (88.45) ***

28 (14.07) 171 (85.93) ***

478 (4.53) 10 069 (95.47)

567 (66.16) 196 (22.87) 94 (10.97)

125 (62.81) 55 (27.64) 19 (9.55)

7892 (74.83) 2016 (19.11) 639 (6.06)

All comparisons of maternal and child’s characteristics between either one of the MMT groups and ND group were analyzed by chi-square test; *P < 0.05, **P < 0.01, ***P < 0.001. a Maternal insured amount was defined on the basis of the NHID. b During the three years before the index child was born. c Within one year after delivery, excluding substance use disorders, within one year after delivery.

in the MMT-related groups were more likely to have discontinuous or no insurance coverage within their first year of life (BM: adjusted odds ratio (aOR) = 1.29, 95% CI: 1.10–1.53; AM: aOR = 1.56, 95% CI: 1.14–2.13). Maternal factors such as maternal insurance status at delivery (aOR = 9.00, 95% CI: 4.81–16.84) and unmarried status (aOR = 1.16, 95% CI: 1.04–1.29) also showed a positive relationship with infants’ incomplete insurance coverage.

89

Table 2 Estimated association linking maternal and child factors with incomplete insurance coverage during the first year of life (N = 11 603). Variables

Incomplete insurance period (0–11 months) Crude OR

Maternal characteristics MMT participation Non-drug exposed 1.00 Before methadone 1.52*** After methadone 1.76*** Maternal insured amount at deliverya Higher insured amount 1.00 0.76 Middle insured amount Near poverty 1.07 1.82 Poverty 8.70*** Non-insured Educational attainment 1.00 College or above 1.02 Middle/high school 1.27* Elementary school Marital status 1.00 Married 1.39*** Unmarried Urbanization of residence 1.00 Urban 0.93 Suburban 0.78* Rural Number of prior deliveriesc ≥2 1.00 1.07 1 0 1.15 Number of prenatal examinations ≥7 1.00 1.19*** 4–6 1–3 1.39*** 0 1.71*** Outpatient visits for mental disordersb No 1.00 1.06 Yes Child’s characteristics Gender Boy Girl Preterm delivery No Yes

aOR

95% CI

1.00 1.29** 1.56**

1.10–1.53 1.14–2.13

1.00 0.79 1.12 1.93* 9.00***

0.52–1.21 0.74–1.70 1.03–3.60 4.81–16.84

1.00 1.05 1.04

0.96–1.15 0.85–1.28

1.00 1.16**

1.04–1.29

1.00 0.89* 0.69***

0.80–0.98 0.56–0.84

1.00 1.09 1.16

0.79–1.51 0.84–1.59

1.00 1.11 1.18* 1.30**

1.00–1.24 1.01–1.39 1.07–1.58

1.00 0.99

0.83–1.17

1.00 1.02

1.00 1.02

0.94–1.11

1.00 1.26*

1.00 1.12

0.93–1.35

OR: odds ratio; aOR: adjusted odds ratio; 95% CI: 95% confidence interval. The association estimations were obtained via simultaneous adjustment for maternal age, delivery year, maternal income, and all listed variables in the logistic regression models. a Maternal insured amount was defined on the basis of the NHID. b Within the year after the delivery, excluding substance use disorders. c During the three years before the index child was born. * P < 0.05. ** P < 0.01. *** P < 0.001.

Frequency of health care visits among the MMT and comparison groups is shown in Fig. 2. In general, the non-drug exposed (ND) matched infants received preventive care and outpatient curative care more frequently than those in the two MMT groups (e.g., percentages of ≥3 visits of preventive care: 59.10% in the ND, 42.22% in the AM, and 35.39% in the BM). As to frequency of ER visits, children in the AM group had a slightly higher proportion in the category “3 or more visits” than those in the ND and BM groups (6.11% vs 4.07% in the ND and 4.37% in the BM), yet the overall group differences were not statistically significant (panel c, P = 0.78 by chi-square test). After simultaneous adjustment for listed variables (see Table 3), we found that BM infants had fewer preventive care visits than matched non-drug exposed (ND) infants (adjusted relative risk [aRR] = 0.85, 95% CI: 0.80–0.90). In contrast, the AM group utilized preventive services as frequently as the matched ND infants,

90

S.-Y. Fang et al. / Drug and Alcohol Dependence 153 (2015) 86–93

Fig. 2. Percentages for number of (a) preventive care, (b) outpatient visit, and (c) emergency room visit during the first year of birth by maternal enrollment in methadone treatment program (N = 10 851).

but had more outpatients and ER visits (aRR = 1.11, 95% CI: 1.01–1.23; aRR = 1.46, 95% CI: 1.11–1.90). Maternal unmarried status (aRR = 0.84, 95% CI: 0.81–0.87) and incomplete insurance coverage (aRR = 0.97, 95% CI: 0.94–0.99) were associated with lower preventive health care utilization within the first year after birth. Additionally, rural residence, postnatal history of mental disorders, higher number of prenatal examination, and preterm delivery were associated with increased utilization of both outpatient and ER services. Children with mothers in poverty increased their ER visits by 109% (95% CI: 1.27–3.43). 4. Discussion Based upon a population-based matched retrospective cohort study, we found that approximately one third of the infants born to mothers in the two MMT groups had incomplete insurance coverage during the first year of life, and such risk was 29–56% higher than the non-drug exposed matched infants. As compared with the non-drug exposed matched infants, infants born before maternal methadone treatment had 15% fewer preventive visits; infants born after maternal enrollment in methadone treatment had an 11% and a 46% increase in outpatient and ER visits, respectively. Our findings demonstrated that women in the methadone treatment program (both BM and AM groups) had a disproportionately

larger number in both lower educational (i.e., elementary school: 11–12% vs 3.7%) and higher educational (e.g., college or above: 61% vs 49%) attainment groups, as compared with the income-matched non-drug exposed group. This might result from two different processes: (i) increased risk of heroin use associated with lower educational attainment, and (ii) job or income loss consequent to heroin use. Important maternal differences between MMT-treated and control groups were also manifested in marital status and postnatal mental health problems, suggesting that children born to heroin- or methadone-using women were more likely to have a disadvantaged family environment. The present study explored the risk of having discontinuous insurance coverage in a high-risk group for health and developmental problems (i.e., infants born to women exposed to heroin or methadone). Prior studies, mostly derived from the U.S., have identified infants’ health insurance coverage as a conceptualized indicator for access to healthcare (Aday and Andersen, 1974; Cummings et al., 2009; Huang et al., 2006; Olson et al., 2005). Our analyses showed that, even with universal health coverage, around 10% of the infants in the MMT group (i.e., both BM and AM) had no insurance coverage in the first year of life. The lack of full insurance coverage among infants born before/after mother’s enrollment in the MMT may be partially attributed to other covariates uncontrolled in this study, such as disadvantaged family condition (e.g.,

S.-Y. Fang et al. / Drug and Alcohol Dependence 153 (2015) 86–93

91

Table 3 Selective factors associated with health care utilization during the first year of life (N = 10 851). Variables

Preventive care visits Crude RRd

aRRd

Outpatient care visits (95% CI)

Maternal characteristics MMT participation (ref = non-drug exposed) 0.70*** 0.85*** 0.80–0.90 Before methadone (BM) 0.77*** 0.97 0.87–1.08 After methadone (AM) Maternal insured amount at delivery (ref = higher insured amount)a Middle insured amount 0.98 0.97 0.87–1.08 0.87** 0.90 0.81–1.00 Near poverty 0.80* 0.88 0.73–1.06 Poverty 0.86 0.99 0.79–1.23 Non-insured Educational attainment (ref = college or above) Middle/high school 1.03* 1.01 0.98–1.03 Elementary school 0.82*** 0.93* 0.87–0.99 Marital status (ref = married) 0.75*** 0.84*** 0.81–0.87 Unmarried Urbanization of residence (ref = urban) 1.01 1.02 0.99–1.05 Suburban 0.88*** 0.93* 0.88–0.99 Rural Number of prior delivery (ref = ≥2)c ** * 1 1.16 1.11 1.01–1.22 0 1.25*** 1.21*** 1.10–1.33 Number of prenatal examination (ref = ≥7) 4–6 0.91*** 0.92*** 0.89–0.95 0.70*** 0.77*** 0.73–0.81 1–3 0.51*** 0.63*** 0.58–0.67 0 Outpatient visits for mental disorder after delivery (ref = no)b Yes 0.99 1.03 0.98–1.08 Child’s characteristics Gender (ref = boy) 1.00 1.00 0.97–1.02 Girl Insurance period within one year after birth (ref = 12 months) *** * 1–11 months 0.95 0.97 0.94–0.99 Preterm delivery (ref = no) 0.86*** 0.99 0.94–1.05 Yes

Crude RRd

aRRd

0.89*** 0.89*

1.00 1.11*

1.23** 1.09 1.11 1.17

Emergency room visit Crude RRe

aRRe

0.95–1.06 1.01–1.23

0.97 1.23

1.05 1.46**

0.91–1.22 1.11–1.90

1.18** 1.09 1.14 1.25

1.05–1.33 0.97–1.23 0.94–1.39 0.99–1.58

1.03 1.00 2.18** 1.01

0.99 1.01 2.09** 1.08

0.70–1.40 0.72–1.42 1.27–3.43 0.56–2.07

1.06*** 0.92*

1.05*** 0.98

1.02–1.07 0.92–1.05

1.05 1.18

1.03 1.18

0.96–1.11 1.00–1.40

0.84***

0.91***

0.88–0.94

0.98

1.01

0.93–1.11

1.13*** 1.18***

1.12*** 1.19***

1.08–1.15 1.13–1.26

0.99 1.25**

0.96 1.21*

0.89–1.05 1.04–1.41

1.12* 1.04

1.10 1.05

1.00–1.20 0.96–1.15

1.24 1.25

1.26 1.23

0.96–1.65 0.93–1.63

1.00 0.85*** 0.66***

0.93*** 0.80*** 0.71***

0.90–0.96 0.76–0.84 0.66–0.75

1.03 0.85* 0.64***

0.92 0.75*** 0.59***

0.84–1.01 0.65–0.87 0.48–0.71

1.19***

1.20***

1.14–1.26

1.29***

1.27***

1.10–1.45

0.89***

0.90***

0.87–0.92

0.95

0.95

0.89–1.02

0.99

1.01

0.98–1.04

1.05

1.02

0.94–1.11

0.99

1.11***

1.05–1.18

1.22**

1.36***

1.16–1.58

(95% CI)

(95% CI)

RR: relative risk; aRR: adjusted relative risk; 95% CI: 95% confidence interval. a Maternal insured amount was defined on the basis of the NHID. b Within one year after delivery, excluding substance use disorders. c During the three years before the index child was born. d Negative binomial regression analyses. e Zero-inflated negative binomial regression analyses. * P < 0.05. ** P < 0.01. *** P < 0.001.

frequent family relocation), lifestyle related to drug-use (e.g., avoiding insurance application out of fear of being arrested in the process), and inadequate health literacy (e.g., eligibility criteria for national health insurance and social welfare; Larson and Halfon, 2010; Miller and Neaigus, 2001; Pati et al., 2010; Simpson et al., 2005). We also found that risk of having no or discontinuous health insurance was similar in the two MMT subgroups regardless of the timing of methadone treatment initiation. This observation may reflect (i) the heterogeneity in women: the AM mothers may have a longer period of heroin addiction (as indicated by older ages), thus, suffered from long-term social consequences of heroin use (e.g., lack of job skills, social isolation, or drug-related criminal records; Hser et al., 1987; Miller and Neaigus, 2001), and (ii) the gap in service provision to women seeking addiction treatment and lack of coordination between health care and social services. Our findings are generally congruent with the earlier research on children’s discontinuous insurance coverage (Holl et al., 1995; Zeni et al., 2008), which indicates that maternal characteristics, such as being single, having lower income, lower educational attainment, and urban residence, were important predictors. Considering that heroin-addicted women often have two or more above-mentioned sociodemographic characteristics, these factors, together with heroin addiction, can restrict their children’s access to health care in the first year of life. Indeed, such increased risk also manifests

in prenatal and perinatal health care seeking behaviors, including having fewer prenatal visits and having no health insurance coverage at delivery. This finding highlights the need for integrating social welfare and healthcare into the substitute treatment for heroin-addicted women of reproductive ages (e.g., job training, placement services, and education; Kumpfer, 1998). For pregnant women with addiction problems, the strategies to reduce barriers in accessing health care should be designed and implemented from antepartum to postpartum periods. Although infants born before/after mother’s enrollment in the MMT had utilized preventive care services less frequently than those born to the non-drug exposed women, once we adjusted for maternal and child’s characteristics (e.g., insurance status), such risk differences remained significant only in children born before mother’s enrollment in methadone treatment. In Taiwan, children’s preventive care (i.e., well-child service) covers developmental assessment (i.e., physical, linguistic, and social domains) and vaccination. If the child is insured, parents can take their child to any hospital, clinic, or local health center for the well-child visits at no cost. The observed under-utilization might be attributed to unmeasured covariates related to heroin use, such as mothers’ unfavorable lifestyle, concern for getting caught, fears of physician’s disclosure, and inadequate health literacy (Hser et al., 1987; Rosenbaum and Murphy, 1981). For infants who were more likely to be exposed to

92

S.-Y. Fang et al. / Drug and Alcohol Dependence 153 (2015) 86–93

opioids prenatally and postnatally, the under-utilization of preventive services may represent missed opportunities not only for early detection and intervention of developmental problems but also for early identification and reporting related to child abuse. This observation highlights the need for obstetricians or midwives to educate mothers about the child’s preventive services in general before discharge (particularly those with fewer prenatal visits) and the need for home visits to newborns and mothers in unfavorable health conditions (e.g., neonate abstinence syndrome and preterm). Our analyses also found that infants born after mother’s enrollment in the MMT had more outpatient and ER visits as compared with those born to non-drug exposed mothers. Given that methadone-exposed infants usually have greater needs for health care (e.g., lower birth weight and growth retardation) due to methadone exposure in utero and maternal unfavorable lifestyle (Dryden et al., 2009; Hulse et al., 1997; Kakko et al., 2008), this finding may not completely contradict earlier findings suggesting that having better access to primary health care may reduce visits to emergency room (Piehl et al., 2000). The increase in health care utilization can be a result of reducing barriers to healthcare associated with maternal MMT enrollment (e.g., the engagement in MMT may be a pathway into services). Additionally, the increased utilization in preventive care may also aid early identification of health/developmental problems, the referral to appropriate care may increase subsequent utilization of outpatient visits. Finally, prenatal examination consistently appeared to be a significant predictor for outpatient and emergency service utilization, underscoring the need to offer health and social services throughout pregnancy, birth, and early childhood to economically disadvantaged and socially marginalized women.

treatment. Our findings indicated that even with universal health insurance, infants born to mothers in the MMT program were at a greater risk for having discontinuous or even no insurance coverage during their first year of life. Infants born before maternal methadone treatment (when the mothers were likely still using heroin) had fewer visits to preventive service; whereas those born after the treatment had more visits to outpatient and emergency room care. Addiction treatment and harm-reduction programs for women of childbearing ages should be delivered in a coordinated way that ensures comprehensiveness and continuity in healthcare, child welfare, and social services from antepartum to postpartum period.

4.1. Limitations and strengths

Acknowledgements

The interpretation of our findings may be hampered by some limitations. First, these analyses were only based on the data of infants (i) who could be cross-linked via the birth-notification system, birth-registration dataset, and national health insurance (85% of children in BNS) and (ii) who survived to the first birthday. The exclusion of infants not in the registry system or died within the first year may underestimate the association between maternal methadone treatment and insurance coverage as well as health care utilization. Additionally, due to the relatively small sample size in two MMT groups, we are restrained to do subgroup analyses (e.g., whether maternal MMT compliance rate affects children’s health service utilization). The generalization of the barriers to health care access associated with BM infants to those born to women with non-treated heroin addiction may be restrained given selective participation in the MMT. Finally, the generalization to other countries and societies is also limited because of variation in health insurance plan, health care system, and addiction treatment program. Despite these limitations, our study has several strengths. First, the linkage of multiple national datasets allowed us to probe some sensitive issues in this high-risk population, which may not otherwise be easily approachable in community or clinical research. The retrospective cohort study building upon administrative data also reduces potential bias associated with differential follow-up process and recall validity. Finally, the relatively larger sample size also permitted this study to examine research questions with adjustment for an array of known important confounders.

We thank those in Taiwan’s Center for Disease Control who collected and managed the methadone registration dataset and the Collaboration Center of Health Information Application (CCHIA), Ministry of Health and Welfare. This work was supported by a grant from the National Health Research Institutes (MDPP04-014 and NHRI-102A1-PDCO-1312141). Dr. Chen was supported by a grant from the Ministry of Education, Aim for the Top University Plan. The grant funders had no role in the design or conduct of the study; the collection, management, analysis, or interpretation of the data; or the preparation, review, or approval of the manuscript.

5. Conclusions This study, to our knowledge, is one of the first few to document the health care accessibility and utilization among infants of heroin-addicted mothers in the methadone maintenance

Role of funding source Nothing declared. Author’s contribution S.Y. Fang performed the data analysis and wrote the first draft of article. S.Y. Huang assisted in the writing of the article and interpretation of the findings. T. Lin and I.K. Ho assisted with the data linkage and reviewed the draft of the article. C.Y. Chen conceptualized the study, framed the analyses, and supervised all stages of study. All the authors have read and approved the final version of manuscript. Conflict of interest None.

References Abdus, S., Selden, T.M., 2013. Adherence with recommended well-child visits has grown, but large gaps persist among various socioeconomic groups. Health Aff. (Millwood) 32, 508–515. Aday, L.A., Andersen, R., 1974. A framework for the study of access to medical care. Health Serv. Res. 9, 208–220. Burns, L., Conroy, E., Mattick, R.P., 2010. Infant mortality among women on a methadone program during pregnancy. Drug Alcohol Rev. 29, 551–556. Burns, L., Mattick, R.P., Cooke, M., 2006. The use of record linkage to examine illicit drug use in pregnancy. Addiction 101, 873–882. Chen, C.Y., Lien, Y.T., Yeh, H.H., Su, L.W., Ho, I.K., 2015. Comparison of adverse obstetric outcomes and maternity hospitalization among heroin-exposed and methadone-treated women in Taiwan. Int. J. Drug Policy 26, 191–198. Chen, L., Lu, H.M., Shih, S.F., Kuo, K.N., Chen, C.L., Huang, L.C., 2010. Poverty related risk for potentially preventable hospitalisations among children in Taiwan. BMC Health Serv. Res. 10, 196. Cleary, B.J., Eogan, M., O’Connell, M.P., Fahey, T., Gallagher, P.J., Clarke, T., White, M.J., McDermott, C., O’Sullivan, A., Carmody, D., Gleeson, J., Murphy, D.J., 2012. Methadone and perinatal outcomes: a prospective cohort study. Addiction 107, 1482–1492. Cummings, J.R., Lavarreda, S.A., Rice, T., Brown, E.R., 2009. The effects of varying periods of uninsurance on children’s access to health care. Pediatrics 123, e411–e418. DeVoe, J.E., Krois, L., Edlund, C., Smith, J., Carlson, N.E., 2008. Uninsured but eligible children: are their parents insured? Recent findings from Oregon. Med. Care 46, 3–8.

S.-Y. Fang et al. / Drug and Alcohol Dependence 153 (2015) 86–93 DeVoe, J.E., Ray, M., Graham, A., 2011. Public health insurance in Oregon: underenrollment of eligible children and parental confusion about children’s enrollment status. Am. J. Public Health 101, 891–898. Dietz, P.M., Rizzo, J.H., England, L.J., Callaghan, W.M., Vesco, K.K., Bruce, F.C., Bulkley, J.E., Sharma, A.J., Hornbrook, M.C., 2012. Early term delivery and health care utilization in the first year of life. J. Pediatr. 161, 234–239, e231. Dietz, P.M., Rizzo, J.H., England, L.J., Callaghan, W.M., Vesco, K.K., Bruce, F.C., Bulkley, J.E., Sharma, A.J., Hornbrook, M.C., 2013. Health care utilization in the first year of life among small- and large- for-gestational age term infants. Matern. Child Health J. 17, 1016–1024. Dryden, C., Young, D., Hepburn, M., Mactier, H., 2009. Maternal methadone use in pregnancy: factors associated with the development of neonatal abstinence syndrome and implications for healthcare resources. BJOG 116, 665–671. Farr, S.L., Dietz, P.M., Rizzo, J.H., Vesco, K.K., Callaghan, W.M., Bruce, F.C., Bulkley, J.E., Hornbrook, M.C., Berg, C.J., 2013. Health care utilisation in the first year of life among infants of mothers with perinatal depression or anxiety. Paediatr. Perinat. Epidemiol. 27, 81–88. Hakim, R.B., Bye, B.V., 2001. Effectiveness of compliance with pediatric preventive care guidelines among Medicaid beneficiaries. Pediatrics 108, 90–97. Holl, J.L., Oh, E.H., Yoo, J., Amsden, L.B., Sohn, M.W., 2012. Effects of welfare and maternal work on recommended preventive care utilization among low-income children. Am. J. Public Health 102, 2274–2279. Holl, J.L., Szilagyi, P.G., Rodewald, L.E., Byrd, R.S., Weitzman, M.L., 1995. Profile of uninsured children in the United States. Arch. Pediatr. Adolesc. Med. 149, 398–406. Hser, Y.I., Anglin, M.D., Booth, M.W., 1987. Sex differences in addict careers. 3. Addiction. Am. J. Drug Alcohol Abuse 13, 231–251. Huang, Z.J., Yu, S.M., Ledsky, R., 2006. Health status and health service access and use among children in U.S. immigrant families. Am. J. Public Health 96, 634–640. Hulse, G.K., Milne, E., English, D.R., Holman, C.D., 1997. The relationship between maternal use of heroin and methadone and infant birth weight. Addiction 92, 1571–1579. Hulse, G.K., Milne, E., English, D.R., Holman, C.D., 1998a. Assessing the relationship between maternal opiate use and antepartum haemorrhage. Addiction 93, 1553–1558. Hulse, G.K., Milne, E., English, D.R., Holman, C.D., 1998b. Assessing the relationship between maternal opiate use and neonatal mortality. Addiction 93, 1033–1042. Johnson, K., Greenough, A., Gerada, C., 2003. Maternal drug use and length of neonatal unit stay. Addiction 98, 785–789. Jones, H.E., Kaltenbach, K., Heil, S.H., Stine, S.M., Coyle, M.G., Arria, A.M., O’Grady, K.E., Selby, P., Martin, P.R., Fischer, G., 2010. Neonatal abstinence syndrome after methadone or buprenorphine exposure. N. Engl. J. Med. 363, 2320–2331. Kakko, J., Heilig, M., Sarman, I., 2008. Buprenorphine and methadone treatment of opiate dependence during pregnancy: comparison of fetal growth and neonatal outcomes in two consecutive case series. Drug Alcohol Depend. 96, 69–78. Kennare, R., Heard, A., Chan, A., 2005. Substance use during pregnancy: risk factors and obstetric and perinatal outcomes in South Australia. Aust. N. Z. J. Obstet. Gynaecol. 45, 220–225.

93

Kogan, M.D., Newacheck, P.W., Blumberg, S.J., Ghandour, R.M., Singh, G.K., Strickland, B.B., van Dyck, P.C., 2010. Underinsurance among children in the United States. N. Engl. J. Med. 363, 841–851. Kumpfer, K.L., 1998. Links Between Prevention and Treatment for Drug-Abusing Women and Their Children. Drug Addiction Research and the Health of Women. National Institute on Drug Abuse, Rockville, MD. Larson, K., Halfon, N., 2010. Family income gradients in the health and health care access of US children. Matern. Child Health J. 14, 332–342. Lin, H.C., Chang, Y.P., Wang, P.W., Wu, H.C., Yen, C.N., Yeh, Y.C., Chung, K.S., Chang, H.C., Yen, C.F., 2013. Gender differences in heroin users receiving methadone maintenance therapy in Taiwan. J. Addict. Dis. 32, 140–149. Liu, C.Y., Hung, Y.T., Chuang, Y.L., 2006. Incorporating development stratification of Taiwan townships into sampling design of large scale health interview survey. J. Health Manag. 4, 1–22. Magnani, R.J., Rice, J.C., Mock, N.B., Abdoh, A.A., Mercer, D.M., Tankari, K., 1996. The impact of primary health care services on under-five mortality in rural Niger. Int. J. Epidemiol. 25, 568–577. Miller, M., Neaigus, A., 2001. Networks, resources and risk among women who use drugs. Soc. Sci. Med. 52, 967–978. Olson, L.M., Tang, S.F., Newacheck, P.W., 2005. Children in the United States with discontinuous health insurance coverage. N. Engl. J. Med. 353, 382–391. Pati, S., Mohamad, Z., Cnaan, A., Kavanagh, J., Shea, J.A., 2010. Influence of maternal health literacy on child participation in social welfare programs: the Philadelphia experience. Am. J. Public Health 100, 1662–1665. Piehl, M.D., Clemens, C.J., Joines, J.D., 2000. “Narrowing the Gap”: decreasing emergency department use by children enrolled in the Medicaid program by improving access to primary care. Arch. Pediatr. Adolesc. Med. 154, 791–795. Quimbo, S., Florentino, J., Peabody, J.W., Shimkhada, R., Panelo, C., Solon, O., 2008. Underutilization of social insurance among the poor: evidence from the Philippines. PLoS ONE 3, e3379. Rosenbaum, M., Murphy, S., 1981. Getting the treatment: recycling women addicts. J. Psychoact. Drugs 13, 1–13. Simpson, L., Owens, P.L., Zodet, M.W., Chevarley, F.M., Dougherty, D., Elixhauser, A., McCormick, M.C., 2005. Health care for children and youth in the United States: annual report on patterns of coverage, utilization, quality, and expenditures by income. Ambul. Pediatr. 5, 6–44. Substance Abuse and Mental Health Services Administration, 2012. Results from the 2011 National Survey on Drug Use and Health: Summary of National Findings. NSDUH Series H-44. Substance Abuse and Mental Health Services Administration, Rockville, MD. Webster, J., Chandler, J., Battistutta, D., 1996. Pregnancy outcomes and health care use: effects of abuse. Am. J. Obstet. Gynecol. 174, 760–767. Zeni, M.B., Thompson, D., Tanaka, K., Zwinkels, T., de Hollander, G., Aizenberg, T., Talaat, W.I.A.W., Hussain, S.M.N.A.S., Karim, J.A., Sciulli, N., 2008. Discontinuous health insurance coverage among US children: vulnerable children and policy implications. Int. J. Interdiscip. Soc. Sci. 3, 196–204.