Utilization Patterns and Net Direct Medical Cost to Medicaid of Irritable Bowel Syndrome

Bradley C. Martin; Rahul Ganguly; Sandhya Pannicker; Feride Frech; Victoria Barghout


Curr Med Res Opin. 2003;19(8) 

In This Article

Patients and Methods

For this retrospective matched case-control study, data were obtained from Medicaid administrative claims databases in California between January 1, 1997, and December 31, 1998, and in North Carolina between May 1, 1996, and April 30, 1998. Database availability guided the selection of these two states for analysis. The claims data sets were found to be consistent with supplied documentation and included all reimbursed claims from providers for services and products rendered to persons eligible for Medicaid benefits in North Carolina and for a random 5% sample of Medicaid beneficiaries from California. Both states offer comprehensive health benefits to enrollees, including outpatient prescription, hospital, physician, outpatient clinic, nursing home, and a host of other ancillary health services. In California, where managed care penetration is significant, recipients enrolled in prepaid health care plans were excluded from the study. In North Carolina, however, 17 IBS and 31 control subjects enrolled in managed care plans were retained.

Adults aged 18 or older were included as IBS case subjects if they had at least one claim with a primary or secondary International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code of irritable colon (564.1) during the baseline year (California: January 1 to December 31, 1997; North Carolina: May 1, 1996, to April 30, 1997). Sub-classifications of IBS patients by diarrhea and constipation predominance were not feasible due to the lack of more detailed ICD-9-CM coding for irritable colon. Control subjects had no claims for irritable colon (564.1), abdominal pain (789.0x), or constipation (564.0) during the baseline year.

To eliminate subjects who might have had GI conditions that were misdiagnosed or miscoded as IBS, we excluded Medicaid recipients from the study if they had any of the following ICD-9-CM diagnosis codes during the 1-year baseline period: any primary or secondary code for malignant neoplasm of digestive organs and peritoneum (150.xx to 159.xx); any primary or secondary code for inflammatory bowel disease (IBD), Crohn's disease, ulcerative colitis, intestinal vascular insufficiency, or noninfectious enteritis and colitis (555.xx to 558.xx); or any primary or secondary code for diverticulosis (562.01, 562.02, 562.03, 562.11, 562.12, or 562.13). We also applied a post hoc criterion to exclude Medicaid recipients with one or more claims for a long-term care admission or a hospital stay longer than 30 days because we determined that long-term care institutions may undercode conditions such as IBS.

Subjects with IBS were matched with controls 1 : 1 for age, sex, race (in California, white/nonwhite/other; in North Carolina, black/nonblack/other), and length (in months) of Medicaid eligibility. This matching procedure identified exact matches on sex and race and then selected the nearest possible control match for the dates of birth and the number of months eligible in each case. In California, control subjects were also matched according to participation in the three available heath care plan types: (1) unrestricted fee for service, (2) primary care case management, or (3) county-operated managed care.

To check the original IBS case definition and to include Medicaid recipients with IBS who were not coded with irritable colon (564.1), we conducted a sensitivity analysis of the case definition in which a set of 'probable IBS' case subjects was identified and included in the study. The probable IBS subjects were Medicaid recipients aged 18 or older who had had at least one claim with both abdominal pain (789.0x) and constipation (564.0) coded on the same claim form during the baseline period or five or more unique visits/claims with a diagnosis of abdominal pain (789.0x) during the baseline period. In addition, at least one of those claims had to include a primary diagnosis of abdominal pain. For each probable case, a control matched for sex, age, race, and eligibility was selected according to the same criteria as those used for the original IBS controls. The IBS and probable IBS cohorts were analyzed separately; full results are reported for the IBS cohort. Sensitivity analyses are presented only when substantial differences in cost between IBS and probable IBS cases were observed.

The primary outcome measures comprised the following: (1) annual net unadjusted and adjusted incremental Medicaid expenditures for IBS, defined as the difference between mean annualized expenditures for IBS cases and controls; and (2) resource utilization, determined by the number of claims per year. The amount paid by Medicaid was used to determine expenditures (payer perspective). For subjects with IBS, the date of the initial IBS claim during the baseline year was termed the index date. Control subjects were assigned the same index date as their matches in the IBS group. Medicaid expenditures, which were defined as the amount paid by Medicaid for each patient, were calculated from claims filed during the 12 months after the index date.

Expenditures were annualized for each recipient by calculation of the per member per year (PMPY) costs. Unadjusted expenditures and those adjusted for comorbidity, Medicare eligibility, age, race, and sex were calculated according to category of service (i.e. inpatient, physician, outpatient clinic, non-inpatient prescription, and other [not otherwise categorized, such as laboratory and durable medical equipment]) and were summed to obtain a net total amount. Statistically significant differences between cases and controls in unadjusted net expenditures were determined by calculating 95% confidence intervals (CIs) for the difference between IBS case and control groups. Multivariate analysis was used in calculating adjusted net incremental expenditures.

Adjusted net incremental expenditures were calculated by means of a weighted least-squares regression model that used robust Huber-White variance estimates to adjust for heteroscedasticity, which is commonly encountered in health care cost data.[18] The regression model was weighted for duration of eligibility by dividing duration of enrollment (in months) by 12 for each recipient so that influence on the regression line increased with duration of eligibility. Each recipient's annualized cost was specified with the following covariates: case-control dummy variable, Medicare eligibility dummy variable, age in years and (age in years),[2] sex dummy variable, race (white/nonwhite), and comorbidity score in the pre-index year. Because age, sex, and race are weak but consistent predictors of health care costs,[19] they were included as covariates in the regression models to increase explanatory power.

The scoring for comorbidity, a consistent and strong predictor of health care costs, was derived from a collection of 30 diagnostic categories (based on ICD-9-CM codes) that have been shown to influence cost.[20] This collection has been validated in an ambulatory Medicaid population.[20] The weights assigned to each comorbid condition were originally derived from models used to predict 1-year expenditures for a Medicaid population. Total comorbidity score - the 1-year predicted Medicaid expenditures based on the summation of weights assigned to each comorbid condition - was calculated for each subject.

Resource utilization was also described according to category of service, and significance of differences between IBS cases and controls was determined by means of the Student t test. Chi-square and Fisher exact tests were used to assess the differences in frequency data between cases and controls. SAS version 8.1[21] and STATA release 6.0[21,22] were used to perform statistical calculations and to create regression models. A sensitivity analysis was conducted by varying the case definition to include probable IBS cases.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.