- Email: [email protected]
Continuity and Quality of Care for Children With Diabetes Who Are Covered by Medicaid
Continuity and Quality of Care for Children With Diabetes Who Are Covered by Medicaid Dimitri A. Christakis, MD, MPH; Chris Feudtner, MD, PhD, MPH; Catherine Pihoker, MD; Frederick A. Connell, MD, MPH Background.—Poor and minority children with Type 1 diabetes mellitus are at increased risk of severe adverse outcomes as a result of their disease. However, little is known about the quality of care that these children receive and which factors are associated with better quality of care. Objectives.—Our objectives were as follows: 1) to describe the utilization of services associated with quality of care for children with Type 1 diabetes mellitus who are covered by Medicaid and 2) to test the hypothesis that increased continuity of primary care is associated with better care for these children. Design.—Retrospective cohort study. Methods.—Washington State Medicaid claims data for 1997 were used to determine what proportion of children with diabetes had 1) an inpatient or outpatient diagnosis of diabetic ketoacidosis (DKA), 2) a glycosylated hemoglobin (HgA1c) level that had been checked, 3) a retinal examination, and 4) thyroid function studies. Continuity of care was quantified using a pre-established index. Results.—Two hundred fifty-two eligible patients were identified. During the observation year, 20% had an outpatient diagnosis of DKA, 6% were admitted with DKA, 43% visited an ophthalmologist, 54% had their HgA1c checked, and 21% had their thyroid function assessed. Children with high continuity of care were less likely to have DKA as an outpatient (0.30 [0.13–0.71]). Children with medium continuity of care and high continuity of care were less likely to be hospitalized for DKA (0.22 [0.05–0.87] and 0.14 [0.03–0.67], respectively). For preventive services utilization, high continuity of care was associated only with an increased likelihood of visiting an ophthalmologist (2.80 [1.08–3.88]). Conclusions.—The quality of care for Medicaid children with diabetes can be substantially improved. Low continuity of primary care is an identifiable risk factor for DKA. KEY WORDS:
continuity of patient care; diabetes; diabetic ketoacidosis; pediatrics
Ambulatory Pediatrics 2001;2:99 103
T
ype 1 diabetes mellitus (DM-T1) is an uncommon but serious disease affecting approximately 2 out of every 1000 children in the United States.1–3 Diabetes is associated with substantial but modifiable morbidity and mortality risks,4–6 and the quality of diabetes care can be assessed with reliable utilization measures.4,5,7,8 Others have found that poor and minority children with diabetes are at increased risk of hospitalization and death as a result of their disease.9–11 Although previous studies have found that regular visits to a subspecialty clinic are associated with improved glycemic control,12,13 we do not know what specific attributes of primary care delivery are associated with improved outcomes. We recently reported that increased continuity of care is associated with decreased preventable hospitalizations and improved immunization rates in children.14–16
Moreover, we found that the association between increased continuity of care and decreased risk of hospitalization was strongest for children with asthma, and we presumed these factors to be secondary to improved medication use.15 In keeping with these previous findings, we hypothesized that consistent contact with a particular provider would be beneficial to patients with diabetes as well. Accordingly, we conducted a study designed to answer 2 questions: What is the quality of care for children with diabetes who are covered by Medicaid? Is continuity of primary care associated with better quality of care for these children? METHODS Patients We analyzed administrative claims data from Washington State Medicaid for 1997. All children younger than 18 years of age at the end of 1997 who had been covered by Medicaid for the entire calendar year of 1997 were eligible for inclusion in the study. We used a single year to ensure completeness and comparability of claims data, but we employed data from the 1992–96 years to identify patients in the cohort. Patients were identified based on the following criteria: they had at least one ICD-9 diagnosis of diabetes (250.0) and at least one pharmacy fill for insulin during the 5 years preceding the observation year.
From the Departments of Pediatrics (Dr Christakis, Dr Feudtner, Dr Pihoker, and Dr Connell) and Health Services (Dr Christakis and Dr Connell), University of Washington, Seattle, Wash.; the Child Health Institute (Dr Christakis, Dr Feudtner, and Dr Connell), Seattle, Wash.; and the Division of Endocrinology (Dr Pihoker), Children’s Hospital, Seattle, Wash. Address correspondence to Dimitri A. Christakis, MD, MPH, Department of Pediatrics, Child Health Institute, University of Washington, 146 N Canal St, Suite 300, Seattle WA 98103-8652 (e-mail: [email protected]). Received for publication June 3, 2000; accepted October 2, 2000. AMBULATORY PEDIATRICS Copyright q 2001 by Ambulatory Pediatric Association
99
Volume 1, Number 2 March-April 2001
100
Christakis et al
AMBULATORY PEDIATRICS
Table 1. Summary of Codes Used to Identify Quality Indicators Indicator Diabetic ketoacidosis Glycosylated hemogloblin Thyroid function
Retinal exams
Code ● ICD-9 250.1 ‘‘diabetic ketoacidosis’’ ● CPT 83036 ‘‘glycohemoglobin’’ ● CPT 84443, 84800 ‘‘thyroid stimulating hormone’’ ● CPT 84435; 84436, 84437, 82756 ‘‘thyroxine’’ ● CPT 80070 ‘‘thyroid panel’’ ● CPT 84480 ‘‘T3, total T3’’ ● CPT 92018, 92012, 92014, 92002, 92004, 92225, 92226 ‘‘ophthalmological exam: new or established patient’’ ● CPT 92499 unlisted ophthalmological service
Quality of Care Indicators We developed 4 quality indicators that could be assessed using claims data. The codes used for these are summarized in Table 1 and are described as follows. Diabetic Ketoacidosis Avoiding diabetic ketoacidosis (DKA) is a goal of diabetes care. Less serious episodes (ie, increased blood glucose and mild ketosis) are often treated on an outpatient basis, whereas more serious episodes require hospitalization. We identified episodes of DKA diagnosed in both outpatient clinics and those associated with hospitalization based on ICD-9 code and place of service. In both cases, DKA had to be the principal diagnosis coded. Importantly, all of the episodes of DKA that we used as our outcomes followed a diagnosis of diabetes, one that had been made during the preceding years, thereby ensuring that these DKA events were not coincident with the initial presentation of DM-T1. Glycosylated Hemoglobin Prospective data have demonstrated that tight control of blood glucose level is associated with improved outcomes in children and adults. 4,5 Glycosylated hemoglobin (HgA1c) is the standard measure of glucose control, one that is used both in short- and long-term studies of diabetes management. The American Diabetes Association guidelines indicate that HgA1c should be checked 4 times a year.6 We defined compliance with this recommendation as having a single claim for HgA1c submitted in the entire observation year. Thyroid Function Studies Because children with diabetes have been shown to be at risk for hypothyroidism, the current recommendations are that thyroid function studies be performed annually.6 Using CPT codes, we ascertained whether this had been accomplished during the observation year. Annual Retinal Exams Evaluation for retinopathy is recommended for all teenagers.6,7 We used CPT codes from billing data to deter-
Table 2. Example of the Continuity of Care Index (COC) Visit sequence*
COC index 1.0 .75 .57 .54 .23 0
AAAAAAAA AAAABAAA ABAABAAA ABAACAAA ABCBAEFA ABCDEFGH
* Each unique letter denotes a different provider.
mine if children between the ages of 13 and 18 years made a visit to an ophthalmologist during the observation year (1997). Although it was not possible to reliably determine the reason for the visit, we presumed that all visits qualified as fulfilling this recommendation. This would introduce a conservative bias by potentially overestimating the frequency of preventive care utilization of this specialty. Continuity Measure Our primary predictor variable was continuity of care. Although continuity of care is a complex construct, a core component of it is consistency of contact between patients and their providers. Several indices for measuring patientprovider continuity of care have been developed.17 As in our previous studies,14–16 we employed the ‘‘continuity of care’’ (COC) index,18 which takes the following general form:
On s
COC 5
2 j
2N
j51
N(N 2 1)
where N 5 total number of visits, nj 5 number of visits to provider j, and s 5 number of providers. The COC takes on values that lie between zero and one. A value of zero signifies maximum dispersion, which occurs when a different provider is seen for every visit. A value of one signifies minimum dispersion, which occurs when the same provider is seen at every visit. Although no ‘‘gold standard’’ of COC exists with which to validate the COC measure, in a randomized controlled trial of increased COC conducted at a veteran’s administration hospital clinic, the COC was used to quantify continuity, and the mean COC value was significantly greater in the treatment group than in the control group.19 To demonstrate the behavior of the COC, several hypothetical patterns, each of which involved 8 visits, are shown in Table 2. Note that as the contacts with providers become more dispersed—from all visits with provider A to every visit with a different provider—the COC moves from one to zero. COC was measured based on claims that had been filed up until the beginning of the observation year (January 1, 1997). Only visits to primary care providers (general pediatricians and family physicians) were included in calculations of patients’ COCs; visits to subspecialists (eg, endocrinologists) were excluded and hence did not affect the values. Because the continuous numeric values associated with
AMBULATORY PEDIATRICS
Quality of Care in Diabetes
101
Table 3. Unadjusted Quality Indicators by Continuity of Care (COC) Tertiles Overall (%)
Low COC (%)
Med COC (%)
High COC (%)
P Value
Untoward events diabetic ketoacidosis (outpatient) diabetic ketoacidosis (hospitalization)
20.2 6.3
28.9 13.3
20.5 3.6
11.9 2.4
,.05 ,.05
Preventive service utilization ophthalmologist visit glycosylated hemoglobin thyroid function
43 54 21
34.9 55.4 18.1
42.2 54.2 19.0
51.2 52.4 27
,.05 .45 .29
the COC have no inherent clinical meaning, we modeled it as low, medium, and high tertiles, thereby creating 3 groups with equal numbers of patients, based on the distribution of values in our sample. Covariates We controlled for gender, age (as of January 1, 1997), days since the diagnosis of diabetes mellitus first appeared in billing data, and non-white race. Children with poorer control may have an increased number of acute visits, which means that making consistent contact with a regular provider is less likely. Were this the case, illness severity might confound the association between poor COC and our outcomes of interest. We therefore included the total number of outpatient primary care visits as a covariate. In addition, we specifically assessed whether the mean number of visits differed among COC tertiles. Statistical Analysis The chi-square test for trend was used for comparing proportions. Independent sample t tests were used for comparing continuous variables. We used logistic regression to model the odds of having each of the quality measures described above during the 1-year observation period. All analyses were conducted using Stata 6.0 (Stata Corporation, College Station, Tex). RESULTS There were 252 patients included in the sample. One half of the patients were male; their mean age was 11.7 years, and the majority (77%) of these males were white. The mean minimal time of disease at the start of the ob-
servation year was 2.87 years. The tertiles for the COC index were as follows: 0–0.18 (low), 0.19–0.32 (medium), and 0.33–1.0 (high). The mean number of visits did not differ among COC tertiles (P 5 .15). Overall, 20% of children had an outpatient diagnosis of DKA in 1997, and 6% were admitted for DKA. During the observation year, 43% of children saw an ophthalmologist, 54% had at least one HgA1c test performed, and 21% had their thyroid function tested. Bivariate Findings Both children with medium and high COC were less likely than those with low COC to have either an outpatient or inpatient diagnosis of DKA (P , .05). In terms of preventive services utilization, children with greater COC were statistically more likely to have an ophthalmologist visit (P , .05)(Table 3). Multivariate Findings The multivariate results are summarized in Table 4. Notably, high COC was associated with a significantly decreased risk of outpatient DKA (odds ratio [OR] 0.30 [0.13–0.71]), and both medium and high continuity were associated with a significantly decreased risk of inpatient DKA (0.22 [0.05–0.87] and 0.14 [0.03–0.67], respectively). High but not medium COC was associated with a significantly increased probability of an ophthalmologist visit (OR 2.80 [1.00–7.89]). There was no association between increased COC and any of the other preventive service utilization measures.
Table 4. Multivariate Model of Quality Indicators for Medicaid Children with Insulin-Dependent Diabetes Mellitus*
Variable Continuity of care low medium high Male Age (years) Days since diagnosis Non–white race Office visits
Diabetes Ketoacidosis (DKA) (Outpt) OR [95% CI]
DKA (Hosp) OR [95% CI]
Ophthalmologist OR [95% CI]
Glycosylated Hemoglobin OR [95% CI]
Thyroid Studies OR [95% CI]
1.0 [Reference] 0.67 [0.32–1.14] 0.30 [0.13–0.71] 0.65 [0.33–1.25] 1.0 [0.93–1.09] 0.99 [0.99–1.00] 1.29 [0.61–2.78] 1.00 [1.00–1.02]
1.0 [Reference] 0.22 [0.05–0.87] 0.14 [0.03–0.67] 1.27 [0.43–3.78] 1.02 [0.89–1.16] 0.99 [0.99–1.00] 2.8 [0.93–9.04] 1.00 [0.99–1.01]
1.0 [Reference] 1.67 [0.86–3.23] 2.80 [1.08–3.88] 0.88 [0.38–2.03] 0.45 [0.37–1.55] 1.0 [0.99–1.00] 0.44 [0.23–1.01] 0.99 [0.99–1.00]
1.0 [Reference] 1.18 [0.52–1.89] 1.22 [0.45–1.57] 0.79 [0.47–1.34] 1.04 [0.97–1.12] 0.99 [0.99–1.00] 0.74 [0.40–1.41] 1.00 [1.00–1.01]
1.0 [Reference] 1.73 [0.80–3.74] 1.03 [0.46–2.30] 1.10 [0.59–2.08] 1.03 [0.96–1.12] 0.99 [0.99–1.00] 0.89 [0.41–1.31] 1.00 [1.00–1.01]
* Statistically significant associations (P , .05) are shown in bold type.
102
Christakis et al
DISCUSSION The overall quality of care for children with insulindependent Type 1 diabetes mellitus who are covered by Medicaid appears to be poor, and COC appears to be associated with a decreased risk of untoward events related to diabetes but not with improved utilization of most preventive services. Without a benchmark comparison group, we cannot know whether the proportion of children with DKA outpatient diagnoses (20%) and with DKA admissions (6%) are higher than could be obtained with optimal care. Others have found that the complication rates are higher for poor and minority children with diabetes—children such as the ones in our sample.9,10,20 However, even in the absence of adequate quality benchmarking, the facts that as few as 54% of children had a HgA1c checked in the observation year, as few as 43% saw an ophthalmologist, and as few as 21% had their thyroid functions tested are concerning. These proportions may be low, in part because providers do not agree with or are unaware of current guideline recommendations. The association of increased COC with decreased risk of complications is consistent with what we found for asthma care.15 The fact that medication compliance is improved in situations in which patients are familiar with their providers21 may in part explain why chronic illness management appears to be especially responsive to the benefits of regular and consistent contact with a provider. Others have reported that regular provider contact is beneficial in children with diabetes.13 Such contact can best be achieved through proactive scheduling and follow-up. If health care delivery systems adopt a reactive stance— come back when there is a problem—it is difficult for them to promote COC with a regular clinician, since acute visits tend to be scheduled based on availability rather than with a child’s designated provider. Of note is the fact that certain trends in the delivery of medical care now lean against such contact. The increasing size of physician groups may lead to a greater dispersion of care amongst available clinicians. The use of ancillary providers, so called physician-extenders, may also dilute regular contact with a specific provider. Although capitation rewards providers who keep their patients healthy, there may be increased pressure to diminish the frequency of visits, since capitation also rewards providers for not seeing patients. Finally, changes in provider networks may sever existing relationships on occasions during which parents are unable to keep their child’s regular pediatrician for reasons related to their insurance coverage. Why might COC be associated with decreased outpatient and inpatient diagnoses of DKA but not with improved preventive service utilization? Perhaps primary providers and patients are more focused on serious, common, more immediate events, such as glycemic control, rather than on more delayed and less common outcomes, such as retinopathy or hypothyroidism). Or perhaps providers are as unfamiliar with the guidelines for the care
AMBULATORY PEDIATRICS
of diabetic children as they are with other guidelines.22 Alternatively, the association between COC and these preventive services may be present, as suggested by the bivariate results, but is perhaps smaller and would therefore require greater statistical power to discern. Finally, poor glycemic control, as potentially manifested by DKA, would likely result in assessment of HgA1c levels, thereby biasing our results toward a finding of no difference between tertiles for this measure. Several limitations of this study warrant mention. First, as with all studies that rely on claims data, we are reliant on the quality of the claims filed and the diagnoses made. For example, we cannot be certain that we identified all children with diabetes, although our criteria are likely to be highly specific, since insulin is prescribed only for this diagnosis. Moreover, the prevalence of diabetes mellitus among Medicaid recipients for our observation year is 1.9 per 1000 children, a rate that is consistent with what others have reported.2,3,23 In addition, some preventive services may have been rendered without a claim being filed, although failure to bill is unlikely to explain all or even a significant proportion of the observed underutilization. The majority of the patients in this study are capitated for primary care. However, with the exception of outpatient diagnosis of DKA, our outcomes are based on referrals, hospitalizations, and laboratory tests, all of which require CPT codes for reimbursement. The second limitation relates to the associations we found with COC. It is possible that parental or patient organization may be associated both with making and keeping regular appointments and with monitoring the disease more closely. Although this potential confounder of ‘conscientiousness’ cannot be measured using our data (and thus cannot be adjusted for in our14–16 or others’ studies12,13), poor COC clearly serves as a marker of children at increased risk for DKA. In conclusion, the overall preventive care for Medicaid children with Type 1 diabetes mellitus can be substantially improved. Moreover, poor COC can be viewed as an identifiable risk factor for untoward events in children with diabetes, since it is associated with a three- to sevenfold increased risk of DKA. This finding can be used by providers and health systems to identify children for whom more intensive intervention or closer surveillance is warranted. ACKNOWLEDGMENTS The authors thank Richard Boyesen and the Washington State Department of Health and Human Services. Dimitri Christakis is a Robert Wood Johnson Generalist Faculty Physician Scholar. Chris Feudtner is a Robert Wood Johnson Clinical Scholar.
REFERENCES 1. Wagenknecht LE, Roseman JM, Alexander WJ. Epidemiology of IDDM in black and white children in Jefferson County, Alabama, 1979–1985. Diabetes. 1989;38:629–633. 2. Gorwitz K, Howen GG, Thompson T. Prevalence of diabetes in Michigan school-age children. Diabetes. 1976;25:122–127. 3. LaPorte RE, Matsushima M. Prevalence and Incidence of Insulin Dependent Diabetes Mellitus. Diabetes in America. Bethesda, Md: National Institutes of Health; 1995:37–47. 4. Effect of intensive diabetes treatment on the development and
AMBULATORY PEDIATRICS
5.
6. 7.
8.
9.
10.
11.
12.
progression of long-term complications in adolescents with insulin-dependent diabetes mellitus: Diabetes Control and Complications Trial. Diabetes Control and Complications Trial Research Group [see comments]. J Pediatr. 1994;125:177–188. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group [see comments]. N Engl J Med. 1993; 329:977–986. ADA Position Statement. Standards of medical care for patients with diabetes mellitus. Diabetes Care. 1998;21:S23–S31. AAP Section in Endocrinology. Screening for retinopathy in the pediatric patient with Type 1 diabetes mellitus. Pediatrics. 1998;101:313–314. DCCT Trial Research Group. Retinopathy and nephropathy in patients with Type 1 diabetes four years after a trial of intensive therapy. N Engl J Med. 2000;342:381–389. Lipton R, Good G, Mikhailov T, Freels S, Donoghue E. Ethnic differences in mortality from insulin-dependent diabetes mellitus among people less than 25 years of age. Pediatrics. 1999; 103:952–956. Kovacs M, Charron-Prochownik D, Obrosky DS. A longitudinal study of biomedical and psychosocial predictors of multiple hospitalizations among young people with insulin-dependent diabetes mellitus. Diabetes Med. 1995;12:142–148. Auslander WF, Thompson S, Dreitzer D, White NH, Santiago JV. Disparity in glycemic control and adherence between African-American and Caucasian youths with diabetes. Diabetes Care. 1997;20:1569–1575. Kaufman FR, Halvorson M, Carpenter S. Association between diabetes control and visits to a multidisciplinary pediatric diabetes clinic. Pediatrics. 1999;103:948–951.
Quality of Care in Diabetes
103
13. Jacobson AM, Hauser ST, Willett J, Wolfsdorf JI, Herman L. Consequences of irregular versus continuous medical follow-up in children and adolescents with insulin-dependent diabetes mellitus. J Pediatr. 1997;131:727–733. 14. Christakis DA, Wright JA, Koepsell TD, Emerson S, Connell FA. Is greater continuity of care associated with less emergency department utilization? Pediatrics. 1999;103:738–742. 15. Christakis DA, Mell L, Koepsell TD, Zimmerman FJ, Connell FA. Continuity of care is associated with decreased emergency department utilization and hospitalization. Pediatrics. In press. 16. Christakis DA, Mell L, Wright J, Davis RL, Connell F. Continuity of care is associated with timely MMR vaccination. AJPH. 2000;90:962–965. 17. Steinwachs DM. Measuring provider continuity in ambulatory care: an assessment of alternative approaches. Med Care. 1979; 17:551–565. 18. Bice TW, Boxerman SB. A quantitative measure of continuity of care. Medical Care. 1977;15:347–349. 19. Wasson JH, Sauvigne AE, Mogielnicki RP, et al. Continuity of outpatient medical care in elderly men. JAMA. 1984;252:2413– 2417. 20. Tull ES, Barinas E. A twofold excess mortality among black compared with white IDDM patients in Allegheny county, Pennsylvania. Diabetes Care. 1996;19:1344–1347. 21. Charney E, Bynum R, Eldredge D, et al. How well do patients take oral penicillin? A collaborative study in private practice. Pediatrics. 1967;40:188–195. 22. Christakis DA, Rivara FP. Pediatricians’ awareness of and attitudes about four clinical practice guidelines. Pediatrics. 1998; 101:825–830. 23. Green A, Gale EA, Patterson CC. Incidence of childhood-onset insulin-dependent diabetes mellitus: the EURODIAB ACE Study. Lancet. 1992;339:905–909.
continuity of patient care; diabetes; diabetic ketoacidosis; pediatrics
Ambulatory Pediatrics 2001;2:99 103
T
ype 1 diabetes mellitus (DM-T1) is an uncommon but serious disease affecting approximately 2 out of every 1000 children in the United States.1–3 Diabetes is associated with substantial but modifiable morbidity and mortality risks,4–6 and the quality of diabetes care can be assessed with reliable utilization measures.4,5,7,8 Others have found that poor and minority children with diabetes are at increased risk of hospitalization and death as a result of their disease.9–11 Although previous studies have found that regular visits to a subspecialty clinic are associated with improved glycemic control,12,13 we do not know what specific attributes of primary care delivery are associated with improved outcomes. We recently reported that increased continuity of care is associated with decreased preventable hospitalizations and improved immunization rates in children.14–16
Moreover, we found that the association between increased continuity of care and decreased risk of hospitalization was strongest for children with asthma, and we presumed these factors to be secondary to improved medication use.15 In keeping with these previous findings, we hypothesized that consistent contact with a particular provider would be beneficial to patients with diabetes as well. Accordingly, we conducted a study designed to answer 2 questions: What is the quality of care for children with diabetes who are covered by Medicaid? Is continuity of primary care associated with better quality of care for these children? METHODS Patients We analyzed administrative claims data from Washington State Medicaid for 1997. All children younger than 18 years of age at the end of 1997 who had been covered by Medicaid for the entire calendar year of 1997 were eligible for inclusion in the study. We used a single year to ensure completeness and comparability of claims data, but we employed data from the 1992–96 years to identify patients in the cohort. Patients were identified based on the following criteria: they had at least one ICD-9 diagnosis of diabetes (250.0) and at least one pharmacy fill for insulin during the 5 years preceding the observation year.
From the Departments of Pediatrics (Dr Christakis, Dr Feudtner, Dr Pihoker, and Dr Connell) and Health Services (Dr Christakis and Dr Connell), University of Washington, Seattle, Wash.; the Child Health Institute (Dr Christakis, Dr Feudtner, and Dr Connell), Seattle, Wash.; and the Division of Endocrinology (Dr Pihoker), Children’s Hospital, Seattle, Wash. Address correspondence to Dimitri A. Christakis, MD, MPH, Department of Pediatrics, Child Health Institute, University of Washington, 146 N Canal St, Suite 300, Seattle WA 98103-8652 (e-mail: [email protected]). Received for publication June 3, 2000; accepted October 2, 2000. AMBULATORY PEDIATRICS Copyright q 2001 by Ambulatory Pediatric Association
99
Volume 1, Number 2 March-April 2001
100
Christakis et al
AMBULATORY PEDIATRICS
Table 1. Summary of Codes Used to Identify Quality Indicators Indicator Diabetic ketoacidosis Glycosylated hemogloblin Thyroid function
Retinal exams
Code ● ICD-9 250.1 ‘‘diabetic ketoacidosis’’ ● CPT 83036 ‘‘glycohemoglobin’’ ● CPT 84443, 84800 ‘‘thyroid stimulating hormone’’ ● CPT 84435; 84436, 84437, 82756 ‘‘thyroxine’’ ● CPT 80070 ‘‘thyroid panel’’ ● CPT 84480 ‘‘T3, total T3’’ ● CPT 92018, 92012, 92014, 92002, 92004, 92225, 92226 ‘‘ophthalmological exam: new or established patient’’ ● CPT 92499 unlisted ophthalmological service
Quality of Care Indicators We developed 4 quality indicators that could be assessed using claims data. The codes used for these are summarized in Table 1 and are described as follows. Diabetic Ketoacidosis Avoiding diabetic ketoacidosis (DKA) is a goal of diabetes care. Less serious episodes (ie, increased blood glucose and mild ketosis) are often treated on an outpatient basis, whereas more serious episodes require hospitalization. We identified episodes of DKA diagnosed in both outpatient clinics and those associated with hospitalization based on ICD-9 code and place of service. In both cases, DKA had to be the principal diagnosis coded. Importantly, all of the episodes of DKA that we used as our outcomes followed a diagnosis of diabetes, one that had been made during the preceding years, thereby ensuring that these DKA events were not coincident with the initial presentation of DM-T1. Glycosylated Hemoglobin Prospective data have demonstrated that tight control of blood glucose level is associated with improved outcomes in children and adults. 4,5 Glycosylated hemoglobin (HgA1c) is the standard measure of glucose control, one that is used both in short- and long-term studies of diabetes management. The American Diabetes Association guidelines indicate that HgA1c should be checked 4 times a year.6 We defined compliance with this recommendation as having a single claim for HgA1c submitted in the entire observation year. Thyroid Function Studies Because children with diabetes have been shown to be at risk for hypothyroidism, the current recommendations are that thyroid function studies be performed annually.6 Using CPT codes, we ascertained whether this had been accomplished during the observation year. Annual Retinal Exams Evaluation for retinopathy is recommended for all teenagers.6,7 We used CPT codes from billing data to deter-
Table 2. Example of the Continuity of Care Index (COC) Visit sequence*
COC index 1.0 .75 .57 .54 .23 0
AAAAAAAA AAAABAAA ABAABAAA ABAACAAA ABCBAEFA ABCDEFGH
* Each unique letter denotes a different provider.
mine if children between the ages of 13 and 18 years made a visit to an ophthalmologist during the observation year (1997). Although it was not possible to reliably determine the reason for the visit, we presumed that all visits qualified as fulfilling this recommendation. This would introduce a conservative bias by potentially overestimating the frequency of preventive care utilization of this specialty. Continuity Measure Our primary predictor variable was continuity of care. Although continuity of care is a complex construct, a core component of it is consistency of contact between patients and their providers. Several indices for measuring patientprovider continuity of care have been developed.17 As in our previous studies,14–16 we employed the ‘‘continuity of care’’ (COC) index,18 which takes the following general form:
On s
COC 5
2 j
2N
j51
N(N 2 1)
where N 5 total number of visits, nj 5 number of visits to provider j, and s 5 number of providers. The COC takes on values that lie between zero and one. A value of zero signifies maximum dispersion, which occurs when a different provider is seen for every visit. A value of one signifies minimum dispersion, which occurs when the same provider is seen at every visit. Although no ‘‘gold standard’’ of COC exists with which to validate the COC measure, in a randomized controlled trial of increased COC conducted at a veteran’s administration hospital clinic, the COC was used to quantify continuity, and the mean COC value was significantly greater in the treatment group than in the control group.19 To demonstrate the behavior of the COC, several hypothetical patterns, each of which involved 8 visits, are shown in Table 2. Note that as the contacts with providers become more dispersed—from all visits with provider A to every visit with a different provider—the COC moves from one to zero. COC was measured based on claims that had been filed up until the beginning of the observation year (January 1, 1997). Only visits to primary care providers (general pediatricians and family physicians) were included in calculations of patients’ COCs; visits to subspecialists (eg, endocrinologists) were excluded and hence did not affect the values. Because the continuous numeric values associated with
AMBULATORY PEDIATRICS
Quality of Care in Diabetes
101
Table 3. Unadjusted Quality Indicators by Continuity of Care (COC) Tertiles Overall (%)
Low COC (%)
Med COC (%)
High COC (%)
P Value
Untoward events diabetic ketoacidosis (outpatient) diabetic ketoacidosis (hospitalization)
20.2 6.3
28.9 13.3
20.5 3.6
11.9 2.4
,.05 ,.05
Preventive service utilization ophthalmologist visit glycosylated hemoglobin thyroid function
43 54 21
34.9 55.4 18.1
42.2 54.2 19.0
51.2 52.4 27
,.05 .45 .29
the COC have no inherent clinical meaning, we modeled it as low, medium, and high tertiles, thereby creating 3 groups with equal numbers of patients, based on the distribution of values in our sample. Covariates We controlled for gender, age (as of January 1, 1997), days since the diagnosis of diabetes mellitus first appeared in billing data, and non-white race. Children with poorer control may have an increased number of acute visits, which means that making consistent contact with a regular provider is less likely. Were this the case, illness severity might confound the association between poor COC and our outcomes of interest. We therefore included the total number of outpatient primary care visits as a covariate. In addition, we specifically assessed whether the mean number of visits differed among COC tertiles. Statistical Analysis The chi-square test for trend was used for comparing proportions. Independent sample t tests were used for comparing continuous variables. We used logistic regression to model the odds of having each of the quality measures described above during the 1-year observation period. All analyses were conducted using Stata 6.0 (Stata Corporation, College Station, Tex). RESULTS There were 252 patients included in the sample. One half of the patients were male; their mean age was 11.7 years, and the majority (77%) of these males were white. The mean minimal time of disease at the start of the ob-
servation year was 2.87 years. The tertiles for the COC index were as follows: 0–0.18 (low), 0.19–0.32 (medium), and 0.33–1.0 (high). The mean number of visits did not differ among COC tertiles (P 5 .15). Overall, 20% of children had an outpatient diagnosis of DKA in 1997, and 6% were admitted for DKA. During the observation year, 43% of children saw an ophthalmologist, 54% had at least one HgA1c test performed, and 21% had their thyroid function tested. Bivariate Findings Both children with medium and high COC were less likely than those with low COC to have either an outpatient or inpatient diagnosis of DKA (P , .05). In terms of preventive services utilization, children with greater COC were statistically more likely to have an ophthalmologist visit (P , .05)(Table 3). Multivariate Findings The multivariate results are summarized in Table 4. Notably, high COC was associated with a significantly decreased risk of outpatient DKA (odds ratio [OR] 0.30 [0.13–0.71]), and both medium and high continuity were associated with a significantly decreased risk of inpatient DKA (0.22 [0.05–0.87] and 0.14 [0.03–0.67], respectively). High but not medium COC was associated with a significantly increased probability of an ophthalmologist visit (OR 2.80 [1.00–7.89]). There was no association between increased COC and any of the other preventive service utilization measures.
Table 4. Multivariate Model of Quality Indicators for Medicaid Children with Insulin-Dependent Diabetes Mellitus*
Variable Continuity of care low medium high Male Age (years) Days since diagnosis Non–white race Office visits
Diabetes Ketoacidosis (DKA) (Outpt) OR [95% CI]
DKA (Hosp) OR [95% CI]
Ophthalmologist OR [95% CI]
Glycosylated Hemoglobin OR [95% CI]
Thyroid Studies OR [95% CI]
1.0 [Reference] 0.67 [0.32–1.14] 0.30 [0.13–0.71] 0.65 [0.33–1.25] 1.0 [0.93–1.09] 0.99 [0.99–1.00] 1.29 [0.61–2.78] 1.00 [1.00–1.02]
1.0 [Reference] 0.22 [0.05–0.87] 0.14 [0.03–0.67] 1.27 [0.43–3.78] 1.02 [0.89–1.16] 0.99 [0.99–1.00] 2.8 [0.93–9.04] 1.00 [0.99–1.01]
1.0 [Reference] 1.67 [0.86–3.23] 2.80 [1.08–3.88] 0.88 [0.38–2.03] 0.45 [0.37–1.55] 1.0 [0.99–1.00] 0.44 [0.23–1.01] 0.99 [0.99–1.00]
1.0 [Reference] 1.18 [0.52–1.89] 1.22 [0.45–1.57] 0.79 [0.47–1.34] 1.04 [0.97–1.12] 0.99 [0.99–1.00] 0.74 [0.40–1.41] 1.00 [1.00–1.01]
1.0 [Reference] 1.73 [0.80–3.74] 1.03 [0.46–2.30] 1.10 [0.59–2.08] 1.03 [0.96–1.12] 0.99 [0.99–1.00] 0.89 [0.41–1.31] 1.00 [1.00–1.01]
* Statistically significant associations (P , .05) are shown in bold type.
102
Christakis et al
DISCUSSION The overall quality of care for children with insulindependent Type 1 diabetes mellitus who are covered by Medicaid appears to be poor, and COC appears to be associated with a decreased risk of untoward events related to diabetes but not with improved utilization of most preventive services. Without a benchmark comparison group, we cannot know whether the proportion of children with DKA outpatient diagnoses (20%) and with DKA admissions (6%) are higher than could be obtained with optimal care. Others have found that the complication rates are higher for poor and minority children with diabetes—children such as the ones in our sample.9,10,20 However, even in the absence of adequate quality benchmarking, the facts that as few as 54% of children had a HgA1c checked in the observation year, as few as 43% saw an ophthalmologist, and as few as 21% had their thyroid functions tested are concerning. These proportions may be low, in part because providers do not agree with or are unaware of current guideline recommendations. The association of increased COC with decreased risk of complications is consistent with what we found for asthma care.15 The fact that medication compliance is improved in situations in which patients are familiar with their providers21 may in part explain why chronic illness management appears to be especially responsive to the benefits of regular and consistent contact with a provider. Others have reported that regular provider contact is beneficial in children with diabetes.13 Such contact can best be achieved through proactive scheduling and follow-up. If health care delivery systems adopt a reactive stance— come back when there is a problem—it is difficult for them to promote COC with a regular clinician, since acute visits tend to be scheduled based on availability rather than with a child’s designated provider. Of note is the fact that certain trends in the delivery of medical care now lean against such contact. The increasing size of physician groups may lead to a greater dispersion of care amongst available clinicians. The use of ancillary providers, so called physician-extenders, may also dilute regular contact with a specific provider. Although capitation rewards providers who keep their patients healthy, there may be increased pressure to diminish the frequency of visits, since capitation also rewards providers for not seeing patients. Finally, changes in provider networks may sever existing relationships on occasions during which parents are unable to keep their child’s regular pediatrician for reasons related to their insurance coverage. Why might COC be associated with decreased outpatient and inpatient diagnoses of DKA but not with improved preventive service utilization? Perhaps primary providers and patients are more focused on serious, common, more immediate events, such as glycemic control, rather than on more delayed and less common outcomes, such as retinopathy or hypothyroidism). Or perhaps providers are as unfamiliar with the guidelines for the care
AMBULATORY PEDIATRICS
of diabetic children as they are with other guidelines.22 Alternatively, the association between COC and these preventive services may be present, as suggested by the bivariate results, but is perhaps smaller and would therefore require greater statistical power to discern. Finally, poor glycemic control, as potentially manifested by DKA, would likely result in assessment of HgA1c levels, thereby biasing our results toward a finding of no difference between tertiles for this measure. Several limitations of this study warrant mention. First, as with all studies that rely on claims data, we are reliant on the quality of the claims filed and the diagnoses made. For example, we cannot be certain that we identified all children with diabetes, although our criteria are likely to be highly specific, since insulin is prescribed only for this diagnosis. Moreover, the prevalence of diabetes mellitus among Medicaid recipients for our observation year is 1.9 per 1000 children, a rate that is consistent with what others have reported.2,3,23 In addition, some preventive services may have been rendered without a claim being filed, although failure to bill is unlikely to explain all or even a significant proportion of the observed underutilization. The majority of the patients in this study are capitated for primary care. However, with the exception of outpatient diagnosis of DKA, our outcomes are based on referrals, hospitalizations, and laboratory tests, all of which require CPT codes for reimbursement. The second limitation relates to the associations we found with COC. It is possible that parental or patient organization may be associated both with making and keeping regular appointments and with monitoring the disease more closely. Although this potential confounder of ‘conscientiousness’ cannot be measured using our data (and thus cannot be adjusted for in our14–16 or others’ studies12,13), poor COC clearly serves as a marker of children at increased risk for DKA. In conclusion, the overall preventive care for Medicaid children with Type 1 diabetes mellitus can be substantially improved. Moreover, poor COC can be viewed as an identifiable risk factor for untoward events in children with diabetes, since it is associated with a three- to sevenfold increased risk of DKA. This finding can be used by providers and health systems to identify children for whom more intensive intervention or closer surveillance is warranted. ACKNOWLEDGMENTS The authors thank Richard Boyesen and the Washington State Department of Health and Human Services. Dimitri Christakis is a Robert Wood Johnson Generalist Faculty Physician Scholar. Chris Feudtner is a Robert Wood Johnson Clinical Scholar.
REFERENCES 1. Wagenknecht LE, Roseman JM, Alexander WJ. Epidemiology of IDDM in black and white children in Jefferson County, Alabama, 1979–1985. Diabetes. 1989;38:629–633. 2. Gorwitz K, Howen GG, Thompson T. Prevalence of diabetes in Michigan school-age children. Diabetes. 1976;25:122–127. 3. LaPorte RE, Matsushima M. Prevalence and Incidence of Insulin Dependent Diabetes Mellitus. Diabetes in America. Bethesda, Md: National Institutes of Health; 1995:37–47. 4. Effect of intensive diabetes treatment on the development and
AMBULATORY PEDIATRICS
5.
6. 7.
8.
9.
10.
11.
12.
progression of long-term complications in adolescents with insulin-dependent diabetes mellitus: Diabetes Control and Complications Trial. Diabetes Control and Complications Trial Research Group [see comments]. J Pediatr. 1994;125:177–188. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group [see comments]. N Engl J Med. 1993; 329:977–986. ADA Position Statement. Standards of medical care for patients with diabetes mellitus. Diabetes Care. 1998;21:S23–S31. AAP Section in Endocrinology. Screening for retinopathy in the pediatric patient with Type 1 diabetes mellitus. Pediatrics. 1998;101:313–314. DCCT Trial Research Group. Retinopathy and nephropathy in patients with Type 1 diabetes four years after a trial of intensive therapy. N Engl J Med. 2000;342:381–389. Lipton R, Good G, Mikhailov T, Freels S, Donoghue E. Ethnic differences in mortality from insulin-dependent diabetes mellitus among people less than 25 years of age. Pediatrics. 1999; 103:952–956. Kovacs M, Charron-Prochownik D, Obrosky DS. A longitudinal study of biomedical and psychosocial predictors of multiple hospitalizations among young people with insulin-dependent diabetes mellitus. Diabetes Med. 1995;12:142–148. Auslander WF, Thompson S, Dreitzer D, White NH, Santiago JV. Disparity in glycemic control and adherence between African-American and Caucasian youths with diabetes. Diabetes Care. 1997;20:1569–1575. Kaufman FR, Halvorson M, Carpenter S. Association between diabetes control and visits to a multidisciplinary pediatric diabetes clinic. Pediatrics. 1999;103:948–951.
Quality of Care in Diabetes
103
13. Jacobson AM, Hauser ST, Willett J, Wolfsdorf JI, Herman L. Consequences of irregular versus continuous medical follow-up in children and adolescents with insulin-dependent diabetes mellitus. J Pediatr. 1997;131:727–733. 14. Christakis DA, Wright JA, Koepsell TD, Emerson S, Connell FA. Is greater continuity of care associated with less emergency department utilization? Pediatrics. 1999;103:738–742. 15. Christakis DA, Mell L, Koepsell TD, Zimmerman FJ, Connell FA. Continuity of care is associated with decreased emergency department utilization and hospitalization. Pediatrics. In press. 16. Christakis DA, Mell L, Wright J, Davis RL, Connell F. Continuity of care is associated with timely MMR vaccination. AJPH. 2000;90:962–965. 17. Steinwachs DM. Measuring provider continuity in ambulatory care: an assessment of alternative approaches. Med Care. 1979; 17:551–565. 18. Bice TW, Boxerman SB. A quantitative measure of continuity of care. Medical Care. 1977;15:347–349. 19. Wasson JH, Sauvigne AE, Mogielnicki RP, et al. Continuity of outpatient medical care in elderly men. JAMA. 1984;252:2413– 2417. 20. Tull ES, Barinas E. A twofold excess mortality among black compared with white IDDM patients in Allegheny county, Pennsylvania. Diabetes Care. 1996;19:1344–1347. 21. Charney E, Bynum R, Eldredge D, et al. How well do patients take oral penicillin? A collaborative study in private practice. Pediatrics. 1967;40:188–195. 22. Christakis DA, Rivara FP. Pediatricians’ awareness of and attitudes about four clinical practice guidelines. Pediatrics. 1998; 101:825–830. 23. Green A, Gale EA, Patterson CC. Incidence of childhood-onset insulin-dependent diabetes mellitus: the EURODIAB ACE Study. Lancet. 1992;339:905–909.