Maternal and Child Health Journal

, Volume 14, Issue 4, pp 580–589

National Disparities in the Quality of a Medical Home for Children

Authors

    • Keck School of MedicineUniversity of Southern California
  • Michael Seid
    • Division of Pulmonary Medicine and Center for Health Care Quality, Division of Health Policy and Clinical EffectivenessCincinnati Children’s Hospital and Medical Center
  • Trevor A. Pickering
    • Keck School of MedicineUniversity of Southern California
  • Kai-Ya Tsai
    • Keck School of MedicineUniversity of Southern California
Article

DOI: 10.1007/s10995-009-0454-5

Cite this article as:
Stevens, G.D., Seid, M., Pickering, T.A. et al. Matern Child Health J (2010) 14: 580. doi:10.1007/s10995-009-0454-5

Abstract

Objectives To examine socio-demographic disparities associated with a quality medical home. Methods A nationally representative sample of children ages 0–17 years (n = 102,353) from the 2003 National Survey of Children’s Health. Risk factors including non-white race/ethnicity, income <200% of the federal poverty level (FPL), uninsured, parent education lesser than high school, and non-English primary household language, were examined in relation to a quality medical home separately and together as a “profile” of risk. Fourteen questions were used to measure five medical home features: access, continuity, comprehensiveness, family-centered care, and coordination. Quality was defined as a value greater than median for each feature and for an overall score. Results Before and after adjustment for child demographics and health status, all studied risk factors were associated with poorer quality medical home features. Uninsured [odds ratio (OR) = 0.43, 95% confidence interval (CI): 0.40–0.47] and low-income children (OR = 0.65, CI: 0.62–0.69) had among the lowest odds of a quality medical home overall and across most features, except coordination that showed an opposite trend. Summarized through risk profiles, children experiencing all five risk factors had 93% lower odds of a quality medical home overall (OR = 0.07, CI: 0.04–0.25) compared to zero risk children. Conclusion This study demonstrates large national disparities in the quality of a medical home for children. That disparities were most prevalent for the uninsured and those in or near poverty, both modifiable risk factors, suggests that reforms to increase coverage and to lift families out of poverty are essential to assuring that children have access to the full complement of appropriate health care services including a quality medical home.

Keywords

Medical homePrimary careQuality of careDisparitiesSES

Introduction

The provision of a medical home for children has become a major focus of pediatric health care practice and policy [1]. The concept of a medical home closely resembles the characteristics of high quality primary care and shares many of its cardinal features. The American Academy of Pediatrics (AAP) emphasizes seven features of a medical home: (1) accessible, (2) continuous, (3) comprehensive, (4) coordinated, (5) family-centered, (6) compassionate, and (7) culturally effective [2]. The first four are characteristics of high quality primary care, and as such have been linked with greater preventive care, fewer emergency department visits, lower health care costs, and better health outcomes [3].

As a health care delivery strategy, a medical home was originally promoted for children with special health care needs (CSHCN) because of the greater necessity for care coordination. The medical home concept has broad policy appeal, however, and has been widely promoted for children generally. Nationally, many pediatric advocacy groups cite the establishment of medical homes for all children as a reason for expanding public health insurance coverage. Healthy People 2010, the nation’s main health and health care agenda setting guide, specifically monitors the extent to which all children have a “specific regular source” of ongoing primary care [4].

Nonetheless, there are likely to be considerable disparities in having a quality medical home for children generally. Previous studies have shown racial and ethnic, socioeconomic, and insurance related disparities in children having a regular source of care, a major component of establishing a medical home [59]. Several studies have also examined similar disparities in the discrete characteristics of primary care for children (e.g., accessibility of care and continuity of the patient–provider relationship) [1012].

Two studies have examined disparities in the prevalence of a medical home in a national sample of CSHCN, but not among children generally. These studies found that half of children received care that met established criteria for all five of the medical home characteristics. The studies also showed that the likelihood of having a quality medical home was lower for racial/ethnic minorities and those in lower socioeconomic status [13, 14]. Such findings have also been replicated in two states [15, 16]. There have been no estimates of the national prevalence of, and disparities in, a medical home for children generally.

The current study uses data from the National Survey of Children’s Health to examine the quality of a medical home for five of the seven features defined by the AAP in a national sample of children. The study aims to determine whether differences exist in having a quality medical home according to a set of demographic risk factors for poor access and quality of care based on race/ethnicity, poverty status, parent education level, insurance status, and language spoken at home. The experiences of those with multiple demographic risk factors are analyzed through a “profile” or index of risk previously developed based on these risk factors.

Methods

Data Source and Sampling

This study uses nationally representative data from the publicly available 2003 National Survey of Children’s Health (NSCH) conducted by the National Center for Health Statistics (NCHS) and the Maternal and Child Health Bureau of the Department of Health and Human Services. The NSCH is a module of the State and Local Area Integrated Telephone Survey. A random-digit-dial sample of households with children under the age of 18 years was selected from each of the 50 States and the District of Columbia. Households were sampled so that about 2,000 interviews were conducted in each state to assure that state-level estimates could be produced. More information on the NSCH design is provided in its methodology report [17]. This study was approved by the USC Office for Protection of Research Subjects.

The survey was conducted from January 2003 to July 2004 and contains 102,353 completed interviews. Households containing at least one child <18 years of age were eligible. One child meeting the criteria was selected at random as the subject of the interview. The adult in the household with the most knowledge about the child’s health and health care was chosen to respond to the interview, and in approximately 96% of the interviews this was one of the child’s parents. A Spanish-speaking interviewer was available if needed, and 5.9% of all interviews were conducted in Spanish.

The survey response rates are calculated using the Council of American Survey Research Organizations (CASRO) guidelines. The resolution rate, indicating the percent of phone numbers that could be identified as residential or non-residential, was 91.6%. The screener completion rate, the percent of known residences reporting whether a child lived in the household, was 87.8%. The interview completion rate was 68.8%, and the overall final response rate (the product of the three rates) was 55.3% [17].

In order to weight the data to reflect the national child population, a sampling weight was assigned to each response to account for several factors. A base weight was obtained, equal to the reciprocal of the probability of the particular phone line being sampled in a given state, and was adjusted for multiple telephone lines in the household, multiple children in the household, and non-response due to unknown household status and unknown household eligibility. This weighting allows for nationally representative estimates to be made of the quality of a medical home.

Measures

Child Risk Factors

This study examines five child risk factors for poor health care access and quality. They are based on parent-reported child race/ethnicity, family poverty status, parent education level, child health insurance coverage, and family language spoken at home. The categories considered to be “risk” include: (1) non-white race/ethnicity, (2) income <200% of the Federal Poverty Level (FPL) as calculated from reported family income and size by NSCH staff, (3) highest household education level lesser than high school, (4) child uninsured status, and (5) not primarily speaking English at home.

Child Risk Profiles

To summarize the risk factors a child experiences and their collective impact on the presence of a quality medical home, the risk factors are combined into an index of risk (or “risk profile”) that is a count of the co-occurring risk factors, in this case ranging from 0 (meaning the child has no risk factors) to 5 (the child has five risk factors). Risk profiles have been used previously to investigate and summarize disparities in health care access and quality [1821].

Medical Home Features

Five of seven features of the AAP medical home definition are measured: accessibility, continuity, comprehensiveness, family-centered care, and coordination of care. The questions included in the NSCH and used in this analysis are based on the work of Bethell et al. [22]. The NSCH also assessed another feature of medical home—culturally effective care—with a question about the availability of an interpreter. This question was only asked of the <3% of the study sample that said they needed an interpreter in the past year, providing insufficient sample size for reliable analysis by risk profile. While very important, this feature was not included in our analysis here.

Each of the five features of a quality medical home was measured using one to four survey questions. The questions that are assigned to reflect each medical home feature, along with their frequencies, are shown in Appendix. Each question was assigned a score from 0 to 100, with 100 reflecting the best possible medical home. For questions containing J + 1 possible ordered responses (with J being the value for the best response and 0 the worst response), a given response j was assigned a score as:
$$ MH = \frac{j}{J} \times 100. $$

For example, if there were four possible responses to a question (e.g., never, sometimes, usually, and always) they were scored as follows: never = 0, sometimes = 33, usually = 67 and always = 100. A summary value for each of the five medical home features was computed as the average of all non-missing values in each feature. A total medical home score was calculated by averaging all non-missing values of each of the features (i.e., an average of the averages). For example, the comprehensiveness feature is based on three questions about receiving needed care. If the response was yes (a score of 100) to receiving all needed medical care, no (a score of 0) to receiving all routine preventive dental care, and missing for receiving all need prescriptions, the child’s score for this feature was based on the average of the two non-missing scores (100 + 0/2 = a score of 50).

The median value for the total medical home score (87.5) was defined as the cutoff value for having a quality medical home for each feature and overall. With no known a priori cutoff for defining quality, using the median value is perhaps the least arbitrary definition. It reflects a rather high standard of quality, however, representing a response between usually (67) or always (100) to the medical home questions. Thus, to meet the criteria for having a “quality” medical home in this study, the features must on average usually or always be present.

The total medical home score was based on at least three of the medical home features-accessibility, continuity, and comprehensiveness-to which all individuals responded. Half of the questions for each feature must have been answered in order to compute an average value for the feature. Some questions, however, were not applicable to certain children and were deemed as legitimate skips that did not count towards this requirement. About 17% of children did not have a personal doctor and thus legitimately skipped questions about family-centered care (and so had no value computed for that feature). Also, an additional 60% of all children had a regular source of care but did not need to see a specialist or need special equipment. These children (the 17% with no regular source of care, and the 60% not needing to see a specialist or require special equipment, a total of 77% of all children in the sample) were not asked questions about coordination.

Covariates

The study includes measures of child age (continuous in years), gender, health status (excellent, very good, good, and fair/poor), presence of an activity limitation (yes/no), parent employment status (someone in the household worked outside the home at least 50 weeks in the past year vs. not), and geographic region (Northeast, Midwest, South, and West).

Analysis

Imputing Missing Data

Efforts were made to minimize missing data on family poverty level (10.3% missing) and race/ethnicity (1.6% missing). To impute the missing data for poverty level, a logistic regression prediction model was created using ancillary variables in the NSCH. The best model correctly predicted poverty for >80% of the existing values in the dataset, and was used to predict the missing poverty data. Since race/ethnicity is not binary, imputation was conducted by assigning a probability to a particular response using ancillary variables in the NSCH. Analyses were conducted using both imputed and non-imputed data and results were similar if not identical.

Univariate Analyses

Analyses, including descriptive statistics, were conducted using Stata10. Survey procedures were utilized for all analyses to account for the complex survey design and sampling. Descriptive statistics are provided for the national child population including child demographics, risk factors and risk profiles. Proportions and standard errors are presented. Weighted frequencies for each question used to measure medical home, along with legitimate skips, are also shown.

Bivariate Analyses

The distribution of medical home scores (both overall and feature-specific) was non-normal, so medical home scores were dichotomized at the median score. The bivariate relationship of child demographics, risk factors and profiles with the proportion at or above the median score is presented with respective standard errors. We also present the bivariate relationship between common combinations of the risk factors (i.e., reflecting >1% of the population) and the proportion of the population with scores above the median with 95% confidence interval bars in Fig. 1.
https://static-content.springer.com/image/art%3A10.1007%2Fs10995-009-0454-5/MediaObjects/10995_2009_454_Fig1_HTML.gif
Fig. 1

Risk profile components and percent with medical home score greater than median. Note: 95% confidence interval bars are shown for each estimate. The proportion of the population with each risk profile is noted at the bottom of each bar. E = parent education lesser than high school, I = Uninsured, L = Non-English primary language, P = poverty status <200% FPL, R = non-white race/ethnicity

Multivariable Analyses

Multiple logistic regression was used to examine the relationship between individual risk factors and the proportion of the population at or above the median for the total medical home score and the five medical home features. Results were adjusted for study covariates, and odds ratios (OR) and 95% confidence intervals (CI) are presented. Identical analyses were completed with risk profiles. In addition to comparing risk profiles to a reference group of zero risk factors, each risk profile was tested for statistical significance against its preceding profile (by changing the prior profile to be the reference group) to examine whether a dose–response relationship was present.

Results

Table 1 shows that nationally, a large proportion of children are at-risk for poor access and quality of care. More than one-third (41.5%) of children live in families with income <200% of FPL. Nationally, 12.8% of children live in a family where English is not primarily spoken as the primary language, 9.0% of children are uninsured, and 7.8% live in a household where no adult has graduated from high school. Many children experience multiple of these risk factors. Over one-quarter (28.1%) has one risk factor, 16.9% has two, 7.6% has three, 4.9% has four, and just 1.3% has all five.
Table 1

Descriptive statistics and bivariate relationship with total medical home score

Variable

n

Percent

Standard error

Percent greater than median

Standard error

P

Child age

    0–5

33,322

32.7

0.26

55.0

0.5

<0.0001

    6–12

37,054

39.0

0.27

54.0

0.5

    13–17

31,977

28.3

0.25

49.7

0.5

Child gender

    Male

52,598

51.1

0.28

52.3

0.4

0.03

    Female

49,755

48.9

0.28

53.5

0.4

Race/ethnicity

    White

70,845

60.5

0.28

59.5

0.3

<0.0001

    Blacka

9,884

14.5

0.21

47.6

0.8

    Latinoa

13,574

17.6

0.23

36.6

0.8

    Othera

8,050

7.4

0.18

48.1

1.3

Poverty status

    <200% of FPLa

33,864

41.5

0.28

41.3

0.5

<0.0001

    >200% of FPL

68,489

58.5

0.28

61.1

0.3

Health insurance status

    No health care coveragea

8,053

9.0

0.17

26.9

0.8

<0.0001

    Any health care coverage

94,300

91.0

0.17

55.4

0.3

Highest household education level

    Lesser than high schoola

4,685

7.8

0.19

29.0

1.1

<0.0001

    High school or 12 year equivalent

21,349

26.4

0.26

45.4

0.6

    Greater than high school

76,319

65.7

0.28

58.6

0.3

Primary language spoken at home

    English

94,380

87.2

0.23

56.4

0.3

<0.0001

    Othera

7,973

12.8

0.23

28.7

0.9

Child health status

    Excellent

65,252

60.9

0.28

57.6

0.3

<0.0001

    Very good

23,903

23.2

0.23

49.9

0.6

    Good

10,712

12.7

0.20

39.7

0.8

    Fair or poor

2,486

3.2

0.11

37.4

1.7

Activity limitation

    No or unknown

96,913

94.4

0.13

53.3

0.3

<0.0001

    Yes

5,440

5.6

0.13

45.3

1.2

Employment status

    Not employed

8,695

10.0

0.19

39.6

0.9

<0.0001

    Adult in household employed 50+ weeks/year

93,658

90.0

0.19

54.4

0.3

Geographic region

    Northeast

18,369

17.5

0.11

61.4

0.6

<0.0001

    Midwest

25,509

23.4

0.11

56.1

0.4

    South

35,084

36.2

0.15

51.2

0.4

    West

23,391

22.9

0.17

45.8

0.8

Child risk profile

    0 risk factors

50,943

41.3

0.26

63.8

0.3

<0.0001

    1 risk factor

29,431

28.1

0.25

55.7

0.5

    2 risk factors

13,591

16.9

0.23

43.3

0.8

    3 risk factors

4,832

7.6

0.18

30.7

1.1

    4 risk factors

2,806

4.9

0.16

23.3

1.3

    5 risk factors

750

1.3

0.08

9.0

1.6

aIndicates which categories of the risk factors were included in the child risk profile

Table 1 also shows that each risk factor is strongly associated with having a quality medical home. For example, being uninsured was associated with a 28.5 percentage point deficit in the proportion of children with a quality medical home (26.8% vs. 55.3% for the insured, P < .0001). Similarly, living in a family with income <200% of FPL was related to a 20% point deficit in the proportion of children with a quality medical home (41.2% vs. 61.0% for those with income >200% of FPL, P < .0001). A clear gradient was present between risk profiles and the proportion of children with a quality medical home. Nearly two-thirds (63.7%) of children with no risk factors had a quality medical home, vs. 9% for children with five risks (P < .0001).

Table 2 shows that after adjustment for child demographics and health status, each of the risk factors remained associated with a lower quality medical home. Non-English primary language and being uninsured had the lowest odds of having a quality medical home (OR = 0.59, CI: 0.53–0.65; and OR = 0.43, CI: 0.40–0.47, respectively). The risk factors were also similarly strongly correlated with each feature of a medical home. Two exceptions were coordination of care, where non-white race/ethnicity and family income <200% FPL were associated with higher odds of quality (OR = 1.29, CI: 1.15–1.45; and OR = 1.28, CI: 1.15–1.42, respectively), and accessibility where non-white race/ethnicity was associated with a higher odds of quality (OR = 1.19; CI: 1.12–1.27). Other risk factors were not significantly associated with coordination of care.
Table 2

Multivariable logistic regression predicting total medical home score and each feature

Odds of score greater than median (95% confidence intervals)

Variable

Total medical home score

Accessibility

Continuity

Comprehensive

Family-centered

Coordination

Risk factors

Non-white (ref = white)

0.80

(0.76, 0.84)

1.19

(1.12, 1.27)

0.59

(0.55, 0.64)

0.82

(0.77, 0.87)

0.75

(0.71, 0.80)

1.29

(1.15, 1.45)

<200% FPL (ref = 200%+)

0.65

(0.62, 0.68)

0.90

(0.85, 0.95)

0.50

(0.46, 0.53)

0.58

(0.55, 0.61)

0.84

(0.79, 0.89)

1.28

(1.15, 1.42)

Uninsured (ref = insured)

0.43

(0.39, 0.47)

0.57

(0.52, 0.62)

0.35

(0.31, 0.38)

0.37

(0.34, 0.41)

0.79

(0.71, 0.87)

0.92

(0.73, 1.17)

Lesser than HS parent education (ref = HS+)

0.71

(0.63, 0.80)

0.96

(0.85, 1.07)

0.69

(0.61, 0.79)

0.80

(0.71, 0.91)

0.82

(0.71, 0.94)

0.98

(0.74, 1.29)

Non-English primary language (ref = Eng)

0.59

(0.53, 0.65)

0.78

(0.70, 0.86)

0.68

(0.60, 0.76)

0.86

(0.77, 0.97)

0.55

(0.49, 0.62)

1.04

(0.80, 1.36)

Child health measures

Health status (ref = excellent)

    Very good

0.84

(0.80, 0.89)

0.86

(0.81, 0.91)

1.01

(0.93, 1.09)

0.97

(0.91, 1.03)

0.73

(0.69, 0.77)

0.87

(0.79, 0.97)

    Good

0.79

(0.73, 0.85)

0.84

(0.78, 0.92)

0.91

(0.82, 1.01)

0.87

(0.79, 0.95)

0.79

(0.72, 0.86)

0.99

(0.86, 1.15)

    Fair/poor

0.85

(0.73, 1.00)

0.72

(0.62, 0.85)

0.98

(0.80, 1.19)

0.83

(0.70, 0.99)

0.90

(0.76, 1.07)

1.10

(0.88, 1.37)

Has activity limitation (ref = No)

0.87

(0.78, 0.97)

1.03

(0.93, 1.15)

1.54

(1.31, 1.80)

1.18

(1.05, 1.33)

0.97

(0.86, 1.08)

0.86

(0.74, 1.00)

Demographics

Age (continuous)

0.98

(0.97, 0.98)

0.96

(0.95, 0.96)

0.97

(0.96, 0.97)

1.13

(1.12, 1.14)

0.97

(0.97, 0.98)

0.95

(0.94, 0.96)

Female (ref = male)

1.05

(1.00, 1.10)

0.96

(0.91, 1.00)

1.03

(0.96, 1.10)

1.04

(0.99, 1.09)

1.02

(0.97, 1.07)

1.02

(0.93, 1.11)

Employed (ref = unemployed)

1.28

(1.17, 1.40)

1.15

(1.05, 1.25)

1.13

(1.01, 1.26)

1.30

(1.18, 1.43)

1.04

(0.94, 1.14)

1.00

(0.85, 1.18)

Region (ref = northeast)

   Midwest

0.77

(0.72, 0.82)

0.60

(0.56, 0.64)

0.75

(0.68, 0.84)

0.80

(0.74, 0.85)

0.91

(0.85, 0.97)

1.13

(1.00, 1.27)

    South

0.73

(0.68, 0.78)

0.61

(0.57, 0.66)

0.68

(0.62, 0.76)

0.75

(0.71, 0.81)

0.89

(0.84, 0.95)

1.12

(0.99, 1.26)

    West

0.64

(0.59, 0.69)

0.51

(0.47, 0.56)

0.62

(0.55, 0.70)

0.73

(0.67, 0.79)

0.89

(0.82, 0.97)

0.98

(0.83, 1.15)

Note: Estimates that were statistically-significant at P < .05 or better were bolded for ease of identification. All estimates are adjusted for the other variables listed in the table using multivariable logistic regression. HS = High School

Most covariates were also associated with the total medical home score and some of the medical home features. Older age and health status (including activity limitation) were associated with lower odds of quality for each feature, except that activity limitation was associated with higher odds of continuity and older age with higher comprehensiveness. Children living in the Northeast were consistently more likely to meet the criteria for quality for most medical home features.

Table 3 shows that after adjustment for child demographics and health status, each increase in risk profile was associated with a decrease in children with a quality medical home. Compared to zero risks, children with one had 25% lower odds (OR = 0.75, CI: 0.78–0.89) and those with five had 93% lower odds (OR = 0.07, CI: 0.04–0.25) of a quality medical home. Each decrease in odds was statistically different from the preceding profile at P < .001. This was found for continuity, comprehensiveness, and family-centered care, and partly consistent for accessibility. Coordination showed an opposite trend of increasing odds of quality with higher risk profiles (e.g., OR = 1.29, CI: 1.16–1.43 for one risk and OR = 1.57, CI: 1.01–2.43 for four risks).
Table 3

Multivariable logistic regression predicting total medical home score and each feature by risk profile

Odds of score greater than median (95% confidence intervals)

Variable

Total medical home score

Accessibility

Continuity

Comprehensive

Family-centered

Coordination

Risk profile (ref = 0)

    1 risk factor

0.75**

(0.71, 0.79)

0.95

(0.90, 1.00)

0.61**

(0.56, 0.66)

0.62**

(0.58, 0.65)

0.85**

(0.80, 0.89)

1.29**

(1.16, 1.43)

    2 risk factors

0.47**

(0.44, 0.51)

0.94

(0.88, 1.02)

0.29**

(0.26, 0.32)

0.43**

(0.40, 0.47)

0.59**

(0.54, 0.63)

1.52+

(1.30, 1.78)

    3 risk factors

0.28**

(0.25, 0.32)

0.81+

(0.72, 0.90)

0.17**

(0.15, 0.19)

0.31**

(0.27, 0.35)

0.42**

(0.37, 0.48)

1.74

(1.30, 2.32)

    4 risk factors

0.20**

(0.17, 0.23)

0.63*

(0.55, 0.73)

0.12**

(0.10, 0.14)

0.27

(0.23, 0.32)

0.27**

(0.22, 0.32)

1.57

(1.01, 2.43)

    5 risk factors

0.07**

(0.05, 0.10)

0.41*

(0.31, 0.54)

0.04**

(0.03, 0.05)

0.13**

(0.09, 0.18)

0.29

(0.18, 0.48)

0.84

(0.29, 2.50)

Note: Estimates that were statistically significant at P < .05 or better compared to the reference group of zero risk factors were bolded for ease of identification. All estimates are adjusted for child health status, activity limitation, age, gender, parent employment status, and geographic region. + P < .05, * P < .01, ** P < .001 for the difference between a given risk profile and its preceding profile (e.g., 5 risk factors vs. 4 risk factors)

Figure 1 shows that certain combinations of risk factors are associated with larger deficits in having a quality medical home. Consistent with findings in the multivariable analyses, having risk profiles that include income <200% FPL and being uninsured were associated with the largest deficits in having a quality medical home within each profile category (e.g., two risks, three risks, etc.). For example, for two risk factor profiles, that combination had the lowest percent of children with a quality medical home (33.9%) and was statistically different (i.e., lacking overlapping 95% CI bars) from three of four other combinations within the two risk factor category.

Discussion

This study finds large disparities in the prevalence of a quality medical home for vulnerable children. The relationship between the number of risk factors a child has and the proportion with a quality medical home reveals a steep dose-response like gradient, with a six-fold difference between the lowest and highest risk children. Lacking health insurance coverage and being low-income (particularly when they are both present) were among the strongest predictors of lacking a quality medical home. As with many other previous studies showing the relationship between these factors and aspects of access, quality, and outcomes of health care [2325], our results suggest that disparities in having a quality medical home would likely be amenable to policies to expand insurance coverage for children and to those that aim to reduce the number of children living in poverty.

Because having a medical home may be particularly important for children with more complex medical conditions, our findings that only one-third (37.4%) of children in fair or poor health status and that less than half (44.9%) of children with an activity limitation had a quality medical home, are very troublesome. Nationally, this translates into an estimated 1.5 million children in fair/poor health and 2.3 million children (data not shown) with an activity limitation who do not have a medical home that usually or always provides the cardinal features.

That coordination of care was found to be better among higher risk children, including non-whites and children living in lower income families, suggests some health care providers may be targeting their efforts at coordinating care to children who perhaps need the most assistance. Higher risk children are more likely to have poorer health and a higher prevalence of activity limitations, and these families face considerable barriers to accessing specialty care services [26, 27]. While we might hope that all children would be able to receive a high level of care coordination, the assistance that providers and office staff may be delivering to vulnerable families, such as scheduling specialty provider visits and following up once the visits have occurred, would seem to be appropriate considering the limited resources and time that offices may have to spend on care coordination. It is alternately possible that more vulnerable respondents may be more likely to perceive a need for specialty care only when staff or providers directly arrange for or coordinate the specialty care.

There are several limitations to this study. First, the data are cross-sectional and do not demonstrate causality between the risk factors and medical home quality. Second, even after weighting the data for non-response, the moderate response rate of 55% does not allow us to rule out the possibility of some selection bias, though the direction of a bias is not clear. Third, even though a medical home is widely promoted there is only moderate predictive validity demonstrated for the total medical home scale. Both accessibility and continuity features have been shown to be predictive of lower emergency department use rates among children [28, 29]. Evidence among CSHCN also suggests that not having a medical home is associated with decreased satisfaction with care, greater unmet health care needs and more emergency visits [3032].

Fourth, studies have measured a medical home in many different ways. While the NSCH uses many of the same measures in other studies, the NSCH measures many of the features of a medical home simultaneously, whereas most other studies focus on only one or two features [33]. Our choice of the median as the cutoff score for indicating a quality medical home might have overestimated the disparities in quality medical home, since the anchor at the median was between “usually” and “always.” It could be argued that “usually” having these characteristics is sufficiently indicative of a quality medical home. However, re-running the analysis using the lowest quartile (i.e., a score below 66, reflecting responses that average less than “usually”) yields very similar patterns, suggesting that these disparities are not an artifact of the cutoff score.

In conclusion, this study demonstrates national disparities in the prevalence of a quality medical home for children. That the disparities are most prevalent for uninsured children and those in or near poverty, modifiable risk factors, suggests that health care reforms to expand the number of children with health insurance coverage and programs to help lift families out of poverty will be essential to ensuring that all children establish and maintain a quality medical home. While these data do not show whether children with SCHIP and Medicaid coverage would gain as much as privately insured children, our study suggests that insurance in general is very important to a high quality medical home. As such, SCHIP and Medicaid, which cover about one-third of children nationally and have uncertain financial futures in many states due to budget deficits [34], should continue to be funded to maintain coverage for vulnerable children who would benefit from having a medical home. Nonetheless, it is clear that the reasons for lacking a quality medical home are multi-factorial, and that equity will likely only be achieved by addressing the fuller complement of risk factors that children and families face.

Acknowledgments

Dr. Stevens and Dr. Seid conceptualized and designed the research study. Mr. Pickering and Ms. Tsai completed the statistical analyses and contributed to the methodology section of the manuscript. Dr. Stevens drafted the majority of the manuscript with revision and input from all of the co-authors. Dr. Stevens had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The authors do not have any financial or other conflicts of interest regarding the results that are presented in this study. This project was funded by the Federal Maternal and Child Health Bureau (Grant #R40MC07844).

Copyright information

© Springer Science+Business Media, LLC 2009