Potentially preventable care: ambulatory care-sensitive pediatric hospitalizations in South Carolina in 1998.

Objective: We examined pediatric hospitalizations to assess personal and community factors affecting potentially preventable ambulatory care-sensitive condition (ACSC) hospitalizations.

Methods: Data came from the South Carolina 1998 Hospital Inpatient Encounter Database, which yielded 10,156 ACSC discharges among 81,808 pediatric hospitalizations. Analyses were performed at three levels: ACSC as a percentage of all hospitalizations, ACSC patients compared with other patients, and county ACSC rates.

Results: Younger, male, and nonwhite children; children with Medicaid insurance coverage; and children living in rural areas, health professional shortage area-designated counties, and poorer counties with fewer heath care resources were more likely to be hospitalized with ACSCs. A high percentage of children living in poverty and an absence of federally qualified community health centers were predictive of high county ACSC rates.

Conclusion: Poverty and the absence of a provider stowing low-income children increase ACSC rates. Monitoring changes in ACSC rates can be a tool for studying the effects of policy change.

Key Words: ambulatory care-sensitive conditions, federal community health centers, pediatric hospitalization 

**********

Ambulatory care-sensitive condition (ACSC) hospitalizations are those that could be averted by adequate primary care. (1-4) Preventable hospitalizations have been used as indicators of access to and adequacy of primary care. (5-7) High ACSC rates may reflect problems in availability, accessibility, or appropriateness of primary care. (1) Barriers are more common in minority populations, leading to racial disparities in access. (8) ACSC hospitalizations increase the monetary and human cost of medical care, as tertiary treatment replaces more cost-effective, less risky primary care. Children have higher rates of ACSCs than adults. (4-9)

ACSC hospitalizations are useful in health system performance analyses. (3) High ACSC rates may reflect inadequate primary care supply, barriers to access, variations in disease prevalence, differences in propensity to seek care, varying physician practice patterns, or maladaptive lifestyle factors. (1-3,6)

ACSCs in the General Population

Poverty, low socioeconomic status, and black race have consistently been found to be predictors of high ACSCs in the United States. (1,2,7,10,11) Having Medicaid or being uninsured has been associated with preventable hospitalizations, even after controlling for race and income. (12) At the community level, ACSCs have been found to be more common in both rural areas (4,13,14) and highly urban areas. (1,14)

ACSCs among Children

A comparison of three urban communities found significant differences across communities in pediatric hospitalizations and length of stay for many medical conditions, including some ACSCs. (15) Among Medicaid children, hospitalization rates for some ACSCs were higher in rural areas, even controlling for medical need. (16) Differential rates may result from varying physician practice patterns or differences in access to or adequacy of primary care. (16) Goodman et al (17) found greater hospitalizations among children living in zip codes with higher per capita bed supply, living less than 30 minutes from a hospital, and living outside of an academic medical center service area. When infant discretionary hospitalizations were compared across residential areas, hospitalization rates were highest among inner-city children, possibly reflecting more serious illness and less effective health services. (18) From a demographic perspective, children had more ACSCs if they were younger, nonwhite, or uninsured, or had Medicaid insurance coverage. (19)

Purpose

A previous study using 1995 South Carolina discharge data revealed higher ACSC rates among nonwhites, low-income patients, rural residents, and those without insurance or a primary care provider (PCP). (9) This study validated ACSCs as a measure of primary care access and underscored the role of PCPs in decreasing preventable hospitalizations. (9,19) The authors found that 34% of all South Carolina pediatric admissions and 20% of hospital charges for children were for ACSCs. (9) Our study continues the exploration of social and health system issues surrounding pediatric ACSCs by identifying vulnerable populations, documenting key ACSC diagnoses, estimating the costs of pediatric ACSCs, and highlighting health service problem areas.

Methods

Data Sources

South Carolina hospitals submit a uniform discharge summary fore1 (UB-92) for every patient to the South Carolina Budget and Control Board (SCBCB). UB-92 data for all children under the age of 18 who were discharged in 1998 were obtained for this study. Approval was obtained from the South Carolina Data Oversight Council and the University of South Carolina Institutional Review Board. All data were deidentified.

A unique number was assigned to each child. This permitted a count of individual patients and discharges. Available information included age in years, sex, race, primary payer, county of residence, total charges, and ACSC diagnosis. The primary International Classification of Diseases, Ninth Revision, discharge diagnosis was used, except for newborn infants, when the secondary diagnosis was considered. (9)

County-level data were compiled from the Census Bureau, the South Carolina Department of Health and Environmental Control, and the SCBCB. County-level data included population by age, race, sex, federally determined whole-county health professional shortage area (HPSA) status, rural status, percentage of children living below the poverty level, median household income, number of emergency departments, number of federally qualified community health centers (FQCHCs), total physicians, and number of pediatricians. "Rural" counties contained no town or city larger than 50,000 people and were not part of a standard metropolitan statistical area. Other counties were classified as urban. The physician-to-population ratio was calculated as the total number of physicians in the county per 10,000 total population. County-level estimates of children's health insurance status came from the SCBCB.

Measures

The outcome of interest was an ACSC discharge diagnosis. The definitions for pediatric ACSCs came from previous work using South Carolina data. (9) Those authors used the criteria developed by Billings et al, (2) as modified by a panel of South Carolina physicians and limited to pediatric ACSCs. (9) This study adds congenital syphilis to the list of potentially avoidable conditions; a complete list of diagnoses is provided in the Appendix. Analyses were conducted at three levels: discharge, patient, and county. At the discharge level, we analyzed ACSC discharges as a proportion of all pediatric discharges and the rates of ACSC discharges across pediatric population subgroups. Patient-level data were analyzed to compare characteristics of children with ACSCs to other children. We used Census estimates for children in each age group, by county, to identify the total population of children. We calculated the number of non-ACSC children within each county by subtracting the number of children with ACSCs from the population estimates. This provided an estimate for all non-ACSC children, including both children hospitalized for non-ACSC disorders (eg, neoplasms) and children not hospitalized at all. At the county level, ACSC hospitalization rates per 10,000 children were calculated for each of South Carolina's 46 counties. Pediatric ACSC costs were estimated on the basis of charges. A charge-cost adjustment factor was not used.

Statistical Analysis

Statistical analysis was performed using the SAS System for Windows version 8 (SAS Institute, Inc., Cary, NC). Bivariate analysis with [chi square] was used to compare ACSC children to children not hospitalized for ACSCs. Linear regression models were used to determine the relationship between county-level characteristics and ACSC hospitalization rates. Findings are reported by age subgroups. The existing database lacked a code differentiating delivery-related hospitalizations from other non-ACSC hospitalizations in children under 1 year of age. Because the number of in-hospital deliveries was large (n = 50,007) and could affect some areas of analysis, we separated this age group from the rest of our analysis when appropriate.

Results

Description of Childhood Hospitalizations

In 1998, there were 81,808 discharges among children aged 0 to 17 years (Table 1), including 50,007 live births. ACSCs accounted for 10,156 discharges, 12.4% of all discharges, and 31.9% of discharges excluding births. Boys accounted for 50.4% of all pediatric discharges and 53.4% of ACSC discharges. Black children accounted for 38.2% of all pediatric discharges but 43.8% of all ACSC discharges.

The most common ACSC diagnoses were bacterial pneumonia (28.7%), asthma (21.4%), dehydration (12.7%), gastroenteritis (9.5%), and convulsions (6.3%) (Table 2). In children under 5, bacterial pneumonia was most common and asthma was second; this pattern was reversed in children aged 5 and older (Table 2).

A total of 9,043 children had at least one ACSC admission in 1998. Over 90% of children with ACSC hospitalizations had only one discharge, 8% had two ACSC discharges, and less than 2% had multiple discharges (maximum = 11). Children with non-ACSC hospitalizations had a similar distribution of repeat admissions.

Overall, the 81,808 pediatric hospitalizations generated $417,024,307 in charges. Medicaid charges were $228,321,122, or 54.8% of the total. Charges associated with the 10, 156 ACSC discharges totaled $42,248,928 (10.1% of all charges), or $4,160 per discharge. Medicaid ACSC hospitalizations amounted to $22,681,626, 9.9% of total Medicaid charges and 53.7% of all ACSC charges. In children under I year of age, ACSC charges totaled $9,811,844, with Medicaid's portion amounting to $6,813,864 (69%). Hospital charges for uninsured children were $10,843,550, with self-pay ACSC charges totaling $1,503,192, or 13.9% of uninsured charges.

Medicaid was the most common payer across all discharges (50.4%), followed by private insurance (36.6%) and other payment systems (9.0%). Only 3.9% of hospitalizations occurred in uninsured patients, and a very small proportion had Medicare (0.1%). Payment patterns for ACSC hospitalizations were similar: Medicaid, (50.3%), followed by private insurance (37.3%), other sources of payment (8.1%), and no insurance (4.3%). The proportion of uninsured ACSC discharges increased with age. Although only 1.5% of all discharges among children younger than 1 were uninsured, 5.1% of discharges among children aged 1 to 17 were uninsured. The proportion of uninsured discharges was highest among children aged 15 to 17, at 7.0%.

ACSC Hospitalizations as a Proportion of All Hospitalizations

ACSCs accounted for 31.9% &non-delivery-related hospitalizations in children aged 0 to 17. ACSCs were proportionately most frequent during the first year of life, accounting for 44.4% of non-delivery-related hospitalizations in children under 1 year of age. ACSCs accounted for 29.4% of discharges among children aged 1 to 17. ACSCs as a proportion of all hospitalizations declined steadily with age, to 7.9% in children aged 15 to 17.

Rural children disproportionately experienced ACSC discharges. Although 40.4% of all hospitalizations occurred in children from rural counties, 46.4% of ACSC hospitalizations involved such children. Similarly, although 13.5% of all hospitalizations occurred in children living in HPSA areas, 18.5% of ACSC hospitalizations occurred in these children.

Analysis on the basis of sex, race, insurance, and county descriptors was limited to children aged 1 to 17. ACSCs accounted for 33.5% of hospitalizations in boys and 25.8% in girls. ACSCs accounted for 28.9% of all discharges for white children, 30.1% for blacks, and 25.9% for other races. Across payers, ACSCs accounted for 26.9% of all hospitalizations in Medicaid children, 32.0% in privately insured children, 29.6% in uninsured children, and 32.4% in children insured by other payers. ACSCs accounted for 31.5% of all hospitalizations to rural children and 27.8% to nonrural children. Finally, 35.2% of all hospitalizations among children in whole-county HPSAs were for ACSCs versus 28.3% in non-HPSA counties.

ACSC Hospitalization Rates by Children's Characteristics

State-level ACSC rates per 10,000 pediatric population were higher among younger children, male children, and nonwhite children (Table 3). Medicaid or privately insured children had higher ACSC rates than uninsured children. Higher ACSC rates were found in children living in counties that were rural, were a whole-county HPSA, had low median household income, had a high percentage of children living in poverty, had at least one emergency department, and lacked an FQCHC. ACSC rates did not vary with overall physician-to-population ratio.

Children with ACSC Hospitalization Compared with Other Children Living in the State

We compared the 9,043 children experiencing ACSC hospitalizations in 1998 to all other children in the state to identify factors characterizing ACSC children (Table 4). The number of children with no ACSC hospitalization was calculated by subtracting the children with an ACSC hospitalization from the total number of children. Children experiencing a hospitalization for non-ACSC reasons were thus classified with nonhospitalized children as "no ACSC." Children under 1 were more likely to be hospitalized for an ACSC than children aged 1 to 17 (4.3% versus 0.8%). In bivariate analysis, children who were male, black, and Medicaid-insured were more likely to experience an ACSC than their counterparts. Children were more likely to have an ACSC if they lived in counties that were rural, HPSA, or poor; that had an emergency department; or that lacked an FQCHC. Children were more likely to have an ACSC if they lived in counties in the 50 to 75% quartile of physician-to-population ratio. Access to a pediatrician in the county was not associated with ACSCs.

County-level ACSC Rates

Across South Carolina's 46 counties, the absolute number of ACSC hospitalizations per county ranged from 5 to 663, with a mean of 221. ACSC rates at the county level ranged from 24.6 to 351.0 per 10,000 children. For white children, rates ranged from no ACSC discharges to 298.9 per 10,000 children. For black children, rates ranged from 36.0 to 377.9 per 10,000.

South Carolina counties differ in socioeconomic characteristics and health service infrastructures. Although median household income averaged $31,404, county-specific values ranged from $22,731 to $44,291. The mean number of physicians per 10,000 population was 14.8, ranging from 0.7 to 55.7. The numbers of emergency departments and FQCHCs per county ranged from 0 to 5, with means of 1.2 and 1.1, respectively.

Multiple linear regression identified variables predictive of county ACSC rates (Table 5). Because only 46 observations were available, a maximum of five independent variables were assessed at any one time to avoid saturating the model. Independent variables were entered into the models on the basis of evidence from the literature or correlation with the outcome of interest. Race was included because ACSC hospitalization rates were higher among black children; the proportion of blacks among all children ranged from 8.2 to 78.5% across counties. Insurance was not used because it was so highly correlated with poverty (r = 0.95). Instead, we used the more inclusive variable of percentage of children living in poverty.

On the basis of stepwise regression, the only variables predictive of county ACSC rates were the percentage of children living below the poverty line and the presence of an FQCHC. Every 1% increase in children living in poverty was predictive of an increase in county ACSC rates of 4.15 ACSC hospitalizations per 10,000 children. The presence of an FQCHC in a county was predictive of a decrease of 58.12 ACSC hospitalizations per 10,000 children (model statistics: [r.sup.2] = 0.23; overall model P = 0.001; significance for both variables = 0.001). Analyses were rerun separately for black and white children, with essentially equivalent results.

Discussion

Children Experiencing ACSC Hospitalizations Our findings support previous studies showing more ACSCs in younger children, black children, male children, low-income children, and rural children. (9,19) Similarly, the most common pediatric ACSC diagnoses were bacterial pneumonia, asthma, dehydration, gastroenteritis, and convulsions. (4,9,19) Higher rates of ACSCs among Medicaid children suggest that insurance alone does not guarantee access to ambulatory care.2"7 In addition, Medicaid children may have increased morbidity due to poor environmental and social conditions.12 Our finding of lower rates of ACSCs in the uninsured is limited by the small number of uninsured in our population (n - 399), which itself may reflect the impact of the State Children's Health Insurance Program expansion of Medicaid enrollment. The average charge per pediatric ACSC hospitalization found in our study ($4,160) was similar to that ($3,863) found by Shiet al 99) after adjusting for inflation.

Community Characteristics and ACSC Hospitalizations

In bivariate analysis, children from counties with emergency departments had higher ACSC rates than children from counties without emergency departments. The association of emergency rooms with increased ACSCs may reflect their role as entry points for admission to hospitals.

The effects of physician-to-population ratios were difficult to interpret and may be idiosyncratic to South Carolina. ACSC rates were high in counties with the lowest ratios and in counties in the 50 to 75% quartile. The lack of a significant bivariate association between access to pediatricians and ACSC rates was surprising. However, growth in the number of pediatricians generally has not resulted in increased rural pediatricians. (20) Most rural office visits involving children are handled by family of general practice physicians.21

Of the personal and community factors examined, only the percentage of children living in poverty and the presence of an FQCHC in the county were predictive of county ACSC rates. At the comity level, poverty is generally associated with lower maternal education, fewer community resources, and other variables that can adversely influence health.

Lower ACSC hospitalization rates in counties with FQCHCs suggest potential policy implications. The presence of an FQCHC decreased the ACSC rate by more than half (from 105.9 to 58.12 per 10,000 discharges, or 54.9%). Our results support those of Epstein,22 who found that medically underserved areas containing an FQCHC had lower discharge rates for preventable hospitalizations than similar areas without an FQCHC. Other research found lower ACSCs and related emergency room visits when Medicaid patients used FQCHCs for the majority of their primary care. (4) FQCHCs are a major source of care for poor, uninsured, and rural populations. (4,23,24) These findings give additional emphasis to the President's goal of creating new and expanded FQCHC clinics in 1,200 communities by the end of fiscal year 2006. Such centers, when in place, should reduce both unnecessary illness among children and unnecessary costs to insurers.

Clinically, over 50% of pediatric ACSCs were caused by two conditions: bacterial pneumonia and asthma. These diagnoses should be targeted for preventive measures, including education, guideline implementation, and immunizations. Hakim and Bye (25) found that compliance with a series of well-child visits in the Medicaid population was associated with decreases in avoidable hospitalizations, regardless of race, poverty level, and health status. (25)

There are several limitations to this study. ACSCs are only a screening tool and do not perfectly measure primary care system needs or access to quality health care. (1,3) Social factors such as substance abuse, lack of a primary care physician or insurance, a history of noncompliance, or lack of home support may necessitate differing admission thresholds. (2,6,26) Another limitation concerns the interpretation of ecologic data. The county factors in this study were not necessarily true for individuals. However, the results of previous studies support census-level and community-level data as reliable proxies for individual-level data. (27-29) In addition, this study was a secondary analysis of data obtained for other research, (30) which did not contain all variables necessary for complete evaluation of clinical status. A generic medical coding category such as "convulsions," for example, could not be sorted by cause or severity. Next, physician capacity was estimated using all physicians, not just primary care physicians. However, some internists and specialists perform pediatric primary care functions, supporting this approach. (31) Research based on administrative databases is subject to coding errors and unreliable data. (12) Specifically, a code identifying normal deliveries would have distinguished them from hospitalizations fur non-ACSC conditions in children younger than 1. Furthermore, the ideal comparison group for an ACSC study would be children with "ACSC-type" medical conditions who were not hospitalized. This information was unavailable, so we used a comparison group of all non-ACSC children, including those hospitalized for non-ACSCs and those not hospitalized at all. Finally, our study is limited to 1 year of data from one southern state. Additional research should establish the validity of our findings over time and across different geographic settings.

Conclusions

Our study continues the analysis of South Carolina's ACSC rates and helps to identify trends in the state since 1995.9 The data suggest some improvement in children's health, with a decrease in total pediatric ACSC charges and a reduced proportion of ACSCs among total hospitalizations. These changes may be attributable to increased access to PCPs through the State Children's Health Insurance Program, but analyses of the period 1999 to 2001 are needed to confirm a downward trend.

Our study identifies target areas and populations for future improvement in children's health. Counties with a greater proportion of younger, nonwhite, and poor children should be targeted to increase available primary care. Specifically, investing in FQCHCs may be a beneficial method of delivering pediatric care.

Study findings add to the growing body of evidence that ACSC rates can be used to monitor the effects of policy changes, especially those targeting underserved areas and populations. Proposed expansion of the FQCHC program, for example, offers a chance for time-series analysis of the effects of increased provider availability on health outcomes among vulnerable children. Correlating changes in the health care system to subsequent changes in ACSC rates6 will help to determine what best fosters children's health. In the end, disparities in medical care will not be prevented without resolution of the broad social inequities and noneconomic barriers that lead to disparities in health status and access to health care. (26)

Appendix

Pediatric ACSC diagnoses                    ICD-9-CM code

Cogenital syphilis             90
Immunization/preventable
  conditions                   033, 037, 045, 320.0, 390, 391
Grand mal status/other
  epileptic convulsions        345
Convulsions                    780.3
Severe ear, nose, and throat
  infections                   382, 462, 463, 465, 472.1
Tuberculosis                   011, 012, 014, 015, 016, 017, 018
Bacterial pneumonia            481, 482.2, 482.3, 482.9, 483, 485, 486
Asthma                         493
Cellulitis                     681, 682.3, 683, 686
Diabetes                       250.1, 250.2, 250.3, 250.8, 250.9, 250.0
Gastroenteritis                558.9
Kidney/urinary infections      590, 599.0, 599.9
Dehydration                    276.5
Iron-deficiency anemia         280.1, 280.8, 280.9
Nutritional deficiencies       260, 261, 262, 268.0, 268.1
Failure to thrive              783.4
Dental conditions              521, 522, 523, 525, 528
Skin grafts with cellulitis    DRG 263, DRG 264

(a) ACSC, ambulatory care-sensitive condition; ICD-9-CM, International
Classification of Diseases, Ninth Revision, Clinical Modification; DRG,
Diagnosis-Related Group.

Table 1. Pediatric hospitalizations in South Carolina in 1998, by
gender, race, insurance, and county of residence (a)

                                          Age < 1 yr

                            Total           Total            ACSC
                           children       discharges       children
Variables                (n = 50,450)    (n = 55,449)    (n = 2,185)

Sex
  Male                      26,146          29,021          1,188
  Female                    24,304          26,428            997
Race
  White                     30,420          33,181          1,188
  Black                     17,922          19,975            937
  Other                      2,102           2,287            611
Insurance
  Medicaid                  24,925          28,264          1,446
  Private                   18,798          20,055            565
  Uninsured                  1,766           1,849             33
County descriptors
  Urban                     30,865          33,788          1,149
  Rural                     19,585          21,661          1,036
  HPSA (rural subset)       (6,229)         (6,987)          (413)

                          Age < 1 yr             Age 1-17 yr

                             ACSC           Total           Total
                          discharges       children       discharges
Variables                (n = 2,417)     (n = 21,445)    (n = 26,359)

Sex
  Male                      1,348            9,815          12,176
  Female                    1,069           11,630          14,183
Race
  White                     1,298           12,001          14,439
  Black                     1,047            8,921          11,296
  Other                        72             521              622
Insurance
  Medicaid                  1,612           10,245          12,992
  Private                     617            8,173           9,887
  Uninsured                    36            1,217           1,347
County descriptors
  Urban                     1,275           12,261          15,000
  Rural                     1,142            9,184          11,359
  HPSA (rural subset)        (455)          (3,344)         (4,054)

                                 Age 1-17 yr

                             ACSC            ACSC
                           children       discharges
Variables                (n = 6,858)     (n = 7,739)

Sex
  Male                      3,630           4,075
  Female                    3,228           3,664
Race
  White                     3,746           4,178
  Black                     2,974           3,400
  Other                       138             161
Insurance
  Medicaid                  3,027           3,495
  Private                   2,833           3,167
  Uninsured                   366             398
County descriptors
  Urban                     3,712           4,165
  Rural                     3,146           3,574
  HPSA (rural subset)      (1,268)         (1,426)

(a) HPSA, health professional shortage area; ACSC, ambulatory
care-sensitive condition.

Table 2. Hospitalization rates for the top five pediatric ACSCs, by
age (a)

                Bacterial
                pneumonia             Asthma             Dehydration

Age (yr)     No.     Rate (b)     No.     Rate (b)     No.     Rate (b)

0-17        2,913     (30.4)     2,174     (22.7)     1,293     (13.5)
< 1           717    (141.7)       287     (56.7)       339     (67.0)
1-17        2,196     (24.2)     1,887     (20.8)       954     (10.5)
1-4         1,392     (68.8)       880     (43.5)       620     (30.6)
5-9           563     (20.5)       572     (20.9)       222      (8.1)
10-14         171      (6.5)       328     (12.4)        80      (3.0)
15-17          70      (4.1)       107      (6.3)        32      (1.9)

             Gastroenteritis       Convulsions

Age (yr)     No.    Rate (b)     No.    Rate (b)

0-17         967     (10.1)      643      (6.7)
< 1          303     (59.9)       96     (19.0)
1-17         664      (7.3)      547      (6.0)
1-4          389     (19.2)      296     (14.6)
5-9          161      (5.9)      135      (4.9)
10-14         72      (2.7)       68      (2.6)
15-17         42      (2.5)       48      (2.8)

(a) ACSCs, ambulatory care-vensitire conditions.

(b) Rates per 10,000 pediatric population.

Table 3. Hospitalization rates for pediatric ambulatory
care-sensitive conditions (a)

Variables                                           Rate (b)

Characteristics of child
  Age (yr)
    0-17                                             105.9
    <1                                               477.7
    1-17                                              85.2
    1-4                                              201.4
    5-9                                               73.1
    10-14                                             39.9
    15-17                                             35.2
  Sex
    Male                                             111.1
    Female                                           100.4
  Race
    White                                             91.9
    Black                                            126.7
  Insurance
    Medicaid                                         331.6
    Private                                           52.5
    Uninsured                                         29.5
County descriptors
  Urban                                               91.1
  Rural                                              130.1
    Among rural counties
      HPSA                                           145.4
      Non-HPSA                                        99.7
  Median household income (c)
    0-25%                                            150.1
    25-50%                                           140.3
    50-75%                                           129.1
    75-100%                                           80.9
  Percentage of children below poverty level (c)
    0-25%                                             91.7
    25-50%                                            95.6
    50-75%                                           123.5
    75-100%                                          154.6
  Resources
    ED in county                                     107.4
    ED not in county                                  81.3
    CHC in county                                     97.6
    CHC not in county                                122.5
    MD ratio (c)
      0-25%                                          128.7
      25-50%                                          96.9
      50-75%                                         137.5
      75-100%                                         96.6

(a) HPSA, health proleesional shortage area, ED, emergency department:
CHC, federally qualified community health center: MD ratio, No. of
medical doctors per 10,000 total population.

(b) Rates per 10,000 pediatric population.

(c) Quartiles based on 46 country observations.

Table 4. ACSC children compared with all other children (a,b)

                                           ACSC             No ACSC

Variables                               No.      %        No.       %

Age (yr)
  < 1                                  2,185    24.2     48,415     5.1
  1-17                                 6,858    75.8    902,042    94.9
  1-4                                  3,614    52.7    198,825    22.0
  5-9                                  1,809    26.4    272,507    30.1
  10-14                                  918    13.4    264,232    29.2
  15-17                                  517     7.5    169,205    18.7
Sex
  Male                                 4,818    53.3    483,482    50.9
  Female                               4,225    46.7    466,975    49.1
Race
  White                                4,934    55.8    590,966    63.0
  Black                                3,911    44.2    346,989    37.0
Insurance (d)
  Medicaid                             4,473    54.1    149,527    14.8
  Private                              3,398    41.1    717,602    70.8
  Uninsured                              399     4.8    146,601    14.5
County descriptors
  Urban                                4,861    53.8    592,089    62.3
  Rural                                4,182    46.2    358,318    37.7
  Median household income (e)
    0-25%                              1,363    15.1    100,727    10.6
    25-50%                             1,403    15.5    111,617    11.7
    50-75%                             2,440    27.0    208,500    21.9
    75-100%                            3,837    42.4    529,563    55.7
  Percentage of children below poverty level (e)
    0-25%                              3,970    43.9    482,480    50.8
    25-50%                             1,404    15.5    160,876    16.9
    50-75%                             2,332    25.8    211,198    22.2
    75-100%                            1,337    14.8     95,853    10.1
  County health resources
    HPSA                               1,681    18.6    127,659    13.4
    Non-HPSA                           7,362    81.4    822,748    86.6
    ED in county                       8,633    95.5    894,147    94.1
    ED not in county                     410     4.5     56,260     5.9
    CHC in county                      5,593    61.8    636,407    67.0
    CHC not in county                  3,450    38.2    314,000    33.0
    Pediatrician in county             8,705    96.3    913,535    96.1
    No pediatrician in county            338     3.7     36,872     3.9
    MD ratio (e)
      0-25%                            1,190    13.2    103,170    10.9
      25-50%                           2,066    22.8    238,174    25.1
      50-75%                           1,649    18.2    131,841    13.9
      75-100%                          4,138    45.8    477,222    50.2

Variables                              [chi square]

Age (yr)
  < 1                                   6520.1 (c)
  1-17
  1-4
  5-9                                   3965.8 (c)
  10-14
  15-17
Sex
  Male                                    20.8 (c)
  Female
Race
  White                                  196.0 (c)
  Black
Insurance (d)
  Medicaid
  Private                               9979.1 (c)
  Uninsured
County descriptors
  Urban                                  278.2 (c)
  Rural
  Median household income (e)
    0-25%
    25-50%                               664.9 (c)
    50-75%
    75-100%
  Percentage of children below poverty level (e)
    0-25%
    25-50%                               340.1 (c)
    50-75%
    75-100%
  County health resources
    HPSA                                 204.3 (c)
    Non-HPSA
    ED in county                          31.0 (c)
    ED not in county
    CHC in county                        105.8 (c)
    CHC not in county
    Pediatrician in county                 0.5
    No pediatrician in county
    MD ratio (e)
      0-25%
      25-50%                             219.2 (c)
      50-75%
      75-100%

(a) HPSA, health professional shortage area, ED, emergency department:
CHC, federally qualified community health center; MD ratio, No. of
medical doctors/total population X 10,000.

(b) This includes children who were hospitalized for a reason other
than an ACSC and children a who were not hospitalized at all,
calculated by subtracting the number of ACSC children, from total
pediatric population.

(c) P < 0.0001.

(d) Insurance information obtained from South Carolina Stale Budget and
Control Board for children under age 18 in the year 1998.

(e) Quartiles based on 16 county observations.

Table 5. Association of variables with county-level ACSC rates (a)

                                                    Beta (P value)

                                                        Model 1
                                                 [R.sup.2] = 0.12 (b)
                                                       (n = 46)
                                                 model P value = 0.051
Variables                                         estimate (P value)

Percent of pediatric population that is black        -227.15 (0.047)
Percent of pediatric population that is male       -2,105.36 (0.158)
Rural/HPSA (c)                                         16.16 (0.261)
Percent below poverty line                              7.92 (0.017)
Presence of ED                                         41.87 (0.23)
Presence of CHC                                           --
MD ratio                                                  --

                                                    Beta (P value)

                                                        Model 2
                                                 [R.sup.2] = 0.21 (b)
                                                       (n = 46)
                                                 model P value = 0.012
Variables                                         estimate (P value)

Percent of pediatric population that is black       -82.30 (0.440)
Percent of pediatric population that is male              --
Rural/HPSA (c)                                            --
Percent below poverty line                            6.14 (0.051)
Presence of ED                                       34.08 (0.308)
Presence of CHC                                     -46.73 (0.027)
MD ratio                                             -0.59 (0.583)

                                                    Beta (P value)

                                                  Model 3 (stepwise)
                                                 [R.sup.2] = 0.23 (b)
                                                       (n = 46)
                                                 model P value = 0.001
Variables                                         estimate (P value)

Percent of pediatric population that is black             --
Percent of pediatric population that is male              --
Rural/HPSA (c)                                            --
Percent below poverty line                            4.15 (0.006)
Presence of ED                                            --
Presence of CHC                                     -58.12 (0.003)
MD ratio                                                  --

(a) HPSA, health professional shortage area, ED, indicator variable for
presence of emergenev department; CHC, indicator variable for presence
of federally qualified communily health center; MD ratio, No, of
medical doctors/total population x 10,000.

(b) Adjusted [R.sup.2] for each linear regression model.

(c) Combined rural and HPSA variable.

Key Points

* The overall rate of ambulatory care-sensitive condition (ACSC) hospitalizations was 105.9 discharges per 10,000 children.

* Younger, male, and nonwhite children; children with Medicaid insurance coverage; and children living in rural areas, health professional shortage area-designated counties, and poorer counties with fewer heath care resources were morn likely to be hospitalized with ACSCs.

* A high percentage of children living in poverty and an absence of federally qualified community health centers were predictive of high county-level ACSC rates. The presence of a federally qualified community health center in a county was associated with a decrease of 58.12 ACSC hospitalizations per 10,000 children.

References

(1.) Schreiber S, Zielinski T. The meaning of ambulatory care sensitive admissions: Urban and rural perspectives. J Rural Health 1997; 13:276-284.

(2.) Billings J, Zeitel L Lukomnik J. et al. Impact of socioeconomic status on hospital use in New York City. Health Aff (Millwood) 1993;12:162 173.

(3.) Brown AD, Goldacre M J, Hicks N, ct al. Hospitalization for ambulatory care-sensitive conditions: A method for comparative access and quality studies using routinely collected statistics. Can J Public Health 2001; 92:155 159.

(4.) Falik M, Needleman J, Wells BL, et al. Ambulatory care sensitive hospitalizations and emergency visits: Experiences of Medicaid patients using federally qualified health centers. Med Care 2001;39:551 561.

(5.) Parchman ML, Culler S. Primary care physicians and avoidable hospitalizations. J Fam Pract 1994;39:123-128.

(6.) Bindman AB, Grambach K, Osmond D, et at. Preventable hospitalizations and access to health care. JAMA 1995:274:305 311.

(7.) Billings J, Andersen GM. Newman LS. Recent findings on preventable hospitalizations. Health Aff (Millwod) 1996;15:239-249.

(8.) Forrest CB, Starfield B. Entry into primary care and continuity: The effects of access. Ant J Public Health 1998;88:1330-1336.

(9.) Shi L, Samuels ME, Pease M, et al. Patient characteristics associated with hospitalizations for ambulatory care sensitive conditions in South Carolina. South Med J 1999;92:989-998.

(10.) Pappas G, Hadden WC, Kozak LJ, et al. Potentially avoidable hospitalizations: Inequalities in rates between US socioeconomic groups. Am J Public Health 1997;87:811 816.

(11.) Parker JD, Schoendorf KC. Variation in hospital discharges for ambulatory care-sensitive conditions among children. Pediatrics 2000;106(4 Suppl):942-948.

(12.) Weissman JS, Gatsonis C, Epstein AM. Rates of avoidable hospitalization by insurance status in Massachusetts and Maryland. JAMA 1992; 268:2388-2394.

(13.) Silver MP, Babitz ME, Magill MK. Ambulatory care sensitive hospitalization rates in the aged Medicare population in Utah, 1990 to 1994: A rural-urban comparison. J Rural Health 1997;13:285 294.

(14.) Culler SD, Parchman ML. Przybylski M. Factors related to potentially preventable hospitalizations among the elderly. Med Care 1998;36:804 817.

(15.) Perrin JM, Homer CJ, Berwick DM, et al. Variations in rates of hospitalization of children in three urban communities. N Engl J Med 1989; 320:1183-1187.

(16.) Connell FA. Day RW, LoGerfo JP. Hospitalization of medicaid children: Analysis of small area variations in admission rates. Am J Public health 1981;71:606 613.

(17.) Goodman DC, Fisher ES, Gittelsohn A, et al. Why are children hospitalized? The role of non-clinical factors in pediatric hospitalizations. Pediatrics 1994;93:896-902.

(18.) McConnochie KM, Roghmann K J, Liptak GS. Socioeconomic variation in discretionary and mandatory hospitalization of infants: An ecologic analysis. Pediatrics 1997;99:774-784.

(19.) Shi L, Lu N. Individual sociodemographic characteristics associated with hospitalization for pediatric ambulatory care sensitive conditions. J Health Cure Poor Underserved 2000;11:373-384.

(20.) Randolph GD. Pathman DE, Trends in the rural-urban distribution of general pediatricians. Pediatrics 2001;107(2):E18.

(21.) Probst JC, Moore CG, Baxley EG, et al. Rural-urban differences in visits to primary care physicians. Fam Med 2002;34:609-615.

(22.) Epstein AJ. The role of public clinics in preventable hospitalizations among vulnerable populations. Health Serv Res 2001;36:405-420.

(23.) Forrest CB, Whelan EM. Primary care safety-net delivery sites in the United States: A comparison of community health centers, hospital outpatient departments, and physicians' offices. JAMA 2000:284:2077-2083.

(24.) Weiss E, Haslanger K, Cantor JC, Accessibility of primary care services in safety net clinics in New York City. Am J Public Health 2001;91: 1240-1245.

(25.) Hakim RB, Bye BV. Effectiveness of compliance with pediatric preventive care guidelines among Medicaid beneficiaries. Pediatrics 2001;108:90-97.

(26.) Komaronry M, Lurie N, Osmond D, et al. Physician practice style and rates of hospitalization for chronic medical conditions. Meal Care 1996; 34:594 609.

(27.) Waitzman N J, Smith KR. Phantom of the area: Poverty-area residence and mortality in the United States. Am J Public Health 1998;88:973-976.

(28.) Krieger N. Overcoming the absence of socioeconomic data in medical records: Validation and application of a census-based methodology. Am J Public Health 1992;82:703-710.

(29.) Casanova C, Starfield B. Hospitalizations of children and access to primary care: A cross-national comparison, Int J Health Serv, 1995:25: 283-94.

(30.) Probst JC, Samuels ME, Shaw TV, et al. The National Health Service Corps and Medicaid inpatient care: Experience in a southern state. South Med.12003;96:773-781.

(31.) Rosenblatt RA, Hart LG, Baldwin LM, et al. The generalist role of specialty physicians: Is there a hidden system of primary care? JAMA 1998;279:1364-1370.

From the Department of Family and Preventive Medicine and Department of Health Administration, Arnold School of Public Health. University of South Carolina School of Medicine, University of South Carolina, Columbia, SC; and College of Medicine, University of Kentucky, Lexington, KY.

Reprint requests to Janice C. Probst, PhD, South Carolina Rural Health Research Center, Arnold School of Public Health, University of South Carolina, 800 Sumter Avenue, Columbia, SC 29208. Email: jprobst@ gwm.sc.cdu

Accepted June 9, 2003. Copyright [c] 2003 by The Southern Medical Association 0038-4348/03/9609-0850