Creating a computerized database from administrative claims data.

The creation of a computerized database from Medicaid administrative claims data for research purposes is described. Researchers should consult with computer experts at their institution before selecting software for data manipulation and conversion. It is essential to have an accurate layout of the file record before attempting to convert raw claims data into data sets or other data formats. The location of data elements within the claim will vary depending on whether the record comes from a provider, an institution, or a pharmacy. Each claim contains a common header, a variable header, and a claim detail section. The difficulty in analyzing data elements within a claim detail lies in locating the starting point of the claim detail section. So that data elements not in character or numeric formats can be converted, the file record layout must describe the exact format of each data element and its COBOL notation. A data element dictionary is necessary for translating data element coding into usable data. Data elements not necessary for any planned analysis must be eliminated. The data are then "cleaned" to remove any denied or reversed claims and claims that contain incomplete or erroneous data. Regardless of the format data are obtained in, an accurate file record layout and a data element dictionary are essential to the conversion of administrative claims data into a computerized database for data analysis and research purposes.     

An algorithm for the identification of undiagnosed COPD cases using administrative claims data

BACKGROUND: Chronic obstructive pulmonary disease (COPD) is a major cause of death in the United States, but most persons who have airflow obstruction have never been diagnosed with lung disease. This undiagnosed COPD negatively affects health status, and COPD patients may have increased health care utilization several years before the initial diagnosis of COPD is made. OBJECTIVE: To investigate whether utilization patterns derived from analysis of administrative claims data using a discriminant function algorithm could be used to identify undiagnosed COPD patients. METHODS: Each patient who had a new diagnosis of COPD during the study period (N = 2,129) was matched to as many as 3 control subjects by age and gender. Controls were assigned an index date that was identical to that of the corresponding case, and then all health care utilization for cases and controls for the 24 months prior to the initial COPD diagnosis was compared using logistic regression models. Factors that were significantly associated with COPD were then entered into a discriminant function algorithm. This algorithm was then validated using a separate patient population. RESULTS: In the main model, 19 utilization characteristics were significantly associated with preclinical COPD, although most of the power of the discriminant function algorithm was concentrated in a few of these factors. The main model was able to identify COPD patients in the validation population of adult subjects aged 40 years and older (N = 41,428), with a sensitivity of 60.5% and specificity of 82.1%, even without having information on the history of tobacco use for the majority of the group. Models developed and tested on only 12 months of utilization data performed similarly. CONCLUSION: Discriminant function algorithms based on health care utilization data can be developed that have sufficient positive predictive value to be used as screening tools to identify individuals at risk for having undiagnosed COPD.     

Health care resources and costs for treating peripheral artery disease in a managed care population: results from analysis of administrative claims data

OBJECTIVE: Peripheral arterial disease (PAD) is associated with high rates of morbidity and mortality and serves as an important marker for advanced systemic atherosclerosis accompanied by symptomatic or asymptomatic ischemia of the coronary, cerebral, and visceral vasculature. There are little published data on the use of health care resources and costs attributable to PAD. The objectives of this study were to evaluate, from a societal perspective, PAD-related health care resource utilization and to determine the total annualized costs and cost components for patients with PAD, with particular attention to the key outcomes of myocardial infarction (MI), transient ischemic attacks (TIA), stroke, and amputations. METHODS: This study examined medical, hospital and outpatient, and pharmacy claims from a large managed care database with dates of service from January 1, 1999, through August 31, 2003. Patients with PAD were identified from claims using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes (primary or secondary codes), ICD-9-CM procedure codes, current procedural terminology (CPT) codes, or by a pharmacy claim for cilostazol or pentoxifylline. The index date for each patient was the first occurrence of either a medical claim for PAD or a pharmacy claim for 1 of the 2 drugs. Patients were required to be a minimum of 18 years old with continuous plan eligibility. The prevalence of PAD in adults in a managed care setting was also determined, as were annual rates for the key outcomes of MI, TIA, stroke, and amputations. Health care resource utilization and costs were calculated for PAD patients after the index date for a period of at least 12 months per patient for medications, outpatient/physician office visits, laboratory/diagnostic procedures, emergency department visits, and hospitalization. Cost was defined as the allowed charge on each administrative claim, including the amount paid by the insurer plus the amount paid by the health plan members (copay, deductible, and coinsurance). RESULTS: Prior to application of exclusion criteria for patients aged 18 years or older and the minimum period of continuous eligibility, the overall prevalence of PAD was 1.18% of the total managed care organization population.s 6.67 million members. The PAD study cohort consisted of 30,561 patients with a mean age of 70.7 years at index. The most common comorbidities identified in the preindex period for these PAD patients included hypertension (67% of patients); metabolic disorders/hypercholesterolemia (57%); heart disease including cardiomyopathy, dysrhythmias, and heart failure (55%); and ischemic heart disease (47%). Over a mean postindex period of 25.2 months (median 23.4 months), the total mean annualized PAD-related cost was $5,955 per patient per year (PPPY). Hospitalizations accounted for the largest component cost category, averaging $4,442 PPPY or 75% of the total annualized PAD-related cost per PAD patient. PAD-related noncoronary procedures averaged $729 PPPY (12.2% of total annual PAD-related costs), and PAD-related medications (including antihypertensives and lipid-lowering therapy) totaled $610 (10.2% of total annual costs), including $313 PPPY for antihypertensives and $207 for lipid-lowering therapy. For the subgroup of 24,075 newly identified PAD patients, 8,479 (35.2%) were hospitalized during an average 25.2 months of follow-up, with the mean time to first hospitalization of 8.9 months. CONCLUSIONS: Approximately 75% of the total PAD-related patient cost in an average of 25 months of follow-up is contributed by hospital costs, and 35% of patients newly diagnosed with PAD experienced a hospitalization in a mean of 8.9 months after the index diagnosis. Based upon mean annual health and member costs of only $313 PPPY for antihypertensives and $207 for lipid-lowering therapy, drug therapy in PAD patients may be underutilized.     

Impact of noncompliance with urate-lowering drug on serum urate and gout-related healthcare costs: administrative claims analysis

ABSTRACT

Objective: To determine the association between allopurinol compliance and serum urate (sUA) level; and examine the association between sUA and gout-related healthcare costs in a large managed care population.

Research design and methods: This retrospective administrative claims analysis examined subjects with gout (≥2 medical claims with ICD-9-CM diagnosis code 274.xx or ≥1 claim with a gout diagnosis and ≥1 pharmacy claim for allopurinol, probenecid, colchicine, or sulfinpyrazone) between January 1, 2002 and March 31, 2004. Each subject was observed during 1-year pre-index and 1-year post-index periods.

Main outcome measures: Outcomes were allopurinol medication possession ratio (MPR) and compliance (MPR?≥?0.80), sUA (mg/dL), and gout-related healthcare costs. ‘Post-allopurinol’ sUA was measured during three periods after the first observed allopurinol fill: 30–89 days; 90–149 days; ≥150 days. A baseline sUA on or before the start of the post-index period was also identified. Outcomes were stratified by post-allopurinol or baseline sUA and compliance. Generalized linear modeling (GLM) regression measured the impact of baseline sUA on gout-related healthcare costs, controlling for demographic and health status variables.

Results: The study sample comprised 18?243 subjects with mean age of 53.9 years. In all, 55% (n?=?10?073) of subjects used allopurinol. There were 1473 (8.1%) subjects with a post-allopurinol sUA and 2438 (13.4%) subjects with a baseline sUA result. Among all subjects with a post-allopurinol sUA, 45.6% were compliant; between 49.3% and 56.8% of compliant subjects had an sUA?<?6.0?mg/dL compared with 22.5–27.8% of non-compliant subjects, depending on the post-allopurinol time period (all p?<?0.001). GLM results showed gout-related costs associated with baseline sUA?≥?6.0 and?<?9.0?mg/dL were 58% higher (95% confidence interval (CI): 1.012 –2.456; p?=?0.044) than were costs for sUA?<?6.0?mg/dL. There was no significant difference in gout-related costs between baseline sUA?<?6.0?mg/dL and ≥9.0?mg/dL.

Conclusions: Analysis revealed an important associations between allopurinol compliance, sUA, and gout-related costs: compliance was positively associated with favorable sUA (<6.0?mg/dL) in unadjusted comparisons. GLM showed that baseline sUA?<?6.0 was inversely associated with gout-related costs relative to baseline sUA?≥?6.0 and <9.0?mg/dL. Nevertheless, a substantial portion of subjects, even compliant ones, did not achieve sUA?<?6.0?mg/dL. These results should be interpreted carefully in light of study limitations, including incomplete laboratory data, the potentially incorrect inference that medications were taken as prescribed, and lack of generalizability from Medicare managed care enrollees to the broader Medicare population.