Accuracy of diagnostic coding for Medicare patients under the prospective-payment system

Adherence to surveillance among patients with superficial bladder cancer

BACKGROUND

Patients diagnosed with superficial bladder cancer who have not undergone total cystectomy are at high risk for recurrence, and bladder surveillance with cystoscopy is recommended for such patients every 3-6 months. We examined the degree to which bladder cancer patients undergo the recommended surveillance procedures and identified patient and primary care provider characteristics associated with nonadherence to these recommendations.

METHODS

We used information obtained from the Surveillance, Epidemiology, and End Results (SEER) Program-Medicare-linked database to identify 6717 patients aged 65 years or older who were diagnosed with superficial bladder cancer from 1992 through 1996 and who survived for at least 3 years after diagnosis but did not have a total cystectomy. We used information obtained from Medicare claims forms to examine the frequency with which these patients had a surveillance examination of the bladder during each of five contiguous 6-month intervals from month 7 to month 36 following diagnosis. We examined characteristics of patients and their physicians that were associated with low-intensity surveillance (defined as having an examination during fewer than two of the five possible follow-up intervals). Logistic regression was used to calculate odds ratios (ORs) and 95% confidence intervals (CIs). All statistical tests were two-sided.

RESULTS

Only 40% of the entire cohort had an examination during all five intervals; 1216 patients (18.1%) had low-intensity surveillance. Patient characteristics that were independently associated with low-intensity surveillance were being age 75 years or older (adjusted OR = 1.54, 95% CI = 1.35 to 1.74), nonwhite (adjusted OR = 1.94, 95% CI = 1.57 to 2.40), and having favorable tumor histology (adjusted OR = 0.59, 95% CI = 0.48 to 0.72 for poorly differentiated versus referent well-differentiated tumor grade) and high comorbidity (adjusted OR = 1.72, 95% CI = 1.30 to 2.27). Residence in an urban area or in a census tract with low median income was also associated with low-intensity surveillance.

CONCLUSIONS

The actual practice of surveillance for patients with superficial bladder cancer differs substantially from the standards recommended in clinical guidelines.

Effects of hospital volume on life expectancy after selected cancer operations in older adults: a decision analysis

BACKGROUND

In addition to lower operative mortality, patients undergoing selected cancer operations at high volume centers have improved longterm survival. We sought to determine the overall effect of hospital volume on life expectancy after cancer surgery.

STUDY DESIGN

We used a Markov decision analysis model to estimate life expectancy for patients undergoing resection for pancreatic, lung, or colon cancer. Model inputs included probabilities of operative mortality and longterm survival. For input data, we examined operative mortality (in-hospital or within 30 days) stratified by volume in over 400,000 patients undergoing resection for these three cancers using the national Medicare database (1994-1999). Risks of late mortality were abstracted from published studies (MEDLINE, 1966 to present) to model the effect of hospital volume on longterm survival. In analysis, we first calculated life expectancy for patients undergoing surgery at very low, low, medium, high, and very high volume hospitals. We then explored the effects of various regionalization strategies.

RESULTS

Life expectancy increased steadily with hospital volume for all three cancers. Life expectancy after pancreatic cancer resection increased linearly with hospital volume: from 1.9 years at very low volume centers to 3.6 years at very high volume centers. For lung cancer, life expectancy ranged from 5.4 to 6.6 years. Increases in life expectancy for colon cancer were not as dramatic: from 6.8 at very low volume hospitals to 7.4 years at very high volume hospitals. Differences in life expectancy across volume strata were largely attributable to differences in longterm survival, not operative mortality. From a policy perspective, regionalizing surgery for colon cancer would produce the greatest overall life-expectancy gains, but it would require moving most patients.

CONCLUSIONS

Patients aged 65 and older with pancreatic, lung, and colon cancer have substantially greater life expectancy after cancer resection at higher volume hospitals. Further work is needed to understand the mechanisms underlying differences in performance across hospitals in cancer care.

Prognostic factors following curative resection for pancreatic adenocarcinoma: a population-based, linked database analysis of 396 patients

OBJECTIVE

To analyze prognostic factors influencing pancreatic cancer survival following curative resection, using prospectively collected, population-based data.

SUMMARY BACKGROUND DATA

Several studies have analyzed the determinants of long-term survival in postresection pancreatic cancer patients, but the majority of these have been single-institutional chart reviews yielding inconsistent results.

METHODS

This retrospective cohort study examined 396 Medicare-eligible patients over age 65 who were diagnosed with nonmetastatic pancreatic adenocarcinoma and who underwent surgical resection with curative intent while residing in one of the 11 Survival, Epidemiology, and End Results (SEER) registries between January 1991 and December 1996. Linked Medicare data provided information on treatment and comorbidity, while linked census tract data supplied sociodemographic characteristics.

RESULTS

Median survival for the overall study population was 17.6 months, with 1- and 3-year survival rates of 60.1% and 34.3%, respectively. Survival appears to be gradually improving over time, concomitant with a rise in the proportion of patients undergoing surgery in teaching centers. Prognostic variables significantly diminishing survival on univariate analysis included African American race, treatment not in a teaching hospital, lack of adjuvant chemoradiation therapy, as well as histopathologic factors that included tumor size larger than 2 cm in diameter, moderate to poor histologic grade, and positive lymph node metastases. Higher socioeconomic status was associated both with an increased likelihood of receiving adjuvant therapy and improved overall survival. Multivariate analyses indicated the strongest predictors of survival were adjuvant combined chemoradiotherapy, small tumors (<2 cm in diameter), negative lymph nodes, well-differentiated histology, undergoing surgery in a teaching hospital, and high socioeconomic status.

CONCLUSIONS

Although biologic characteristics remain important predictors of survival for patients with resected pancreatic cancer, the most powerful determinant is postoperative adjuvant chemoradiation therapy. An interesting finding that warrants further investigation is the effect of socioeconomic status on both the likelihood of receiving adjuvant treatment and subsequent survival, indicating a possible relationship between the quality of care delivered and outcomes.

Prognostic factors following curative resection for pancreatic adenocarcinoma: a population-based, linked database analysis of 396 patients

OBJECTIVE

To analyze prognostic factors influencing pancreatic cancer survival following curative resection, using prospectively collected, population-based data.

SUMMARY BACKGROUND DATA

Several studies have analyzed the determinants of long-term survival in postresection pancreatic cancer patients, but the majority of these have been single-institutional chart reviews yielding inconsistent results.

METHODS

This retrospective cohort study examined 396 Medicare-eligible patients over age 65 who were diagnosed with nonmetastatic pancreatic adenocarcinoma and who underwent surgical resection with curative intent while residing in one of the 11 Survival, Epidemiology, and End Results (SEER) registries between January 1991 and December 1996. Linked Medicare data provided information on treatment and comorbidity, while linked census tract data supplied sociodemographic characteristics.

RESULTS

Median survival for the overall study population was 17.6 months, with 1- and 3-year survival rates of 60.1% and 34.3%, respectively. Survival appears to be gradually improving over time, concomitant with a rise in the proportion of patients undergoing surgery in teaching centers. Prognostic variables significantly diminishing survival on univariate analysis included African American race, treatment not in a teaching hospital, lack of adjuvant chemoradiation therapy, as well as histopathologic factors that included tumor size larger than 2 cm in diameter, moderate to poor histologic grade, and positive lymph node metastases. Higher socioeconomic status was associated both with an increased likelihood of receiving adjuvant therapy and improved overall survival. Multivariate analyses indicated the strongest predictors of survival were adjuvant combined chemoradiotherapy, small tumors (<2 cm in diameter), negative lymph nodes, well-differentiated histology, undergoing surgery in a teaching hospital, and high socioeconomic status.

CONCLUSIONS

Although biologic characteristics remain important predictors of survival for patients with resected pancreatic cancer, the most powerful determinant is postoperative adjuvant chemoradiation therapy. An interesting finding that warrants further investigation is the effect of socioeconomic status on both the likelihood of receiving adjuvant treatment and subsequent survival, indicating a possible relationship between the quality of care delivered and outcomes.

Administrative data versus corrected administrative data

The purpose of this research was to provide insight into the use of existing administrative data and to identify changes that could be made to improve broad-based use of administrative data. Data were collected on patients hospitalized with pneumonia at a 715 bed hospital in North Carolina in 1996-1997. Patients were selected from administrative databases via diagnosis and charge codes. Outcome variables were length of stay and total hospital charges. Explanatory variables were age, sex, race, insurance type, season of year, admission source (emergency department or other), comorbidity score, care path designation, physician specialty and teaching appointment. These data were collected from administrative data and then from a limited chart review to correct the administrative data. We found no significant differences in economic outcomes between the administrative data and the corrected administrative data. Administrative data appear to be a reliable and cost-effective data source for quality assessment.

Hospitalized avascular necrosis after renal transplantation in the United States

BACKGROUND

The national incidence of and risk factors for hospitalized avascular necrosis (AVN) in renal transplant recipients has not been reported.

METHODS

This historical cohort study consisted of 42,096 renal transplant recipients enrolled in the United States Renal Data System (USRDS) between 1 July 1994 and 30 June 1998. The data source was USRDS files through May 2000. Associations with hospitalizations for a primary diagnosis of AVN (ICD-9 codes 733.4x) within three years after renal transplant were assessed in an intention-to-treat design by Cox regression analysis.

RESULTS

Recipients had a cumulative incidence of 7.1 episodes/1000 person-years from 1994 to 1998. The two-year incidence of AVN did not change significantly over time. Eighty-nine percent of the cases of AVN were due to AVN of the hip (733.42) and 60.2% of patients with AVN underwent total hip arthroplasty (THA); these percentages did not change significantly over time. In the Cox regression analysis, an earlier year of transplant, African American race [adjusted hazard ratio (AHR), 1.65, 95% confidence interval (CI) 1.33 to 2.03], allograft rejection (AHR 1.67, 95% CI 1.35 to 2.07), peritoneal dialysis (vs. hemodialysis; AHR 1.44, 95% CI 1.15 to 1.81), and diabetes (AHR 0.41, 95% CI 0.27 to 0.64) were the only factors independently associated with hospitalizations for AVN.

CONCLUSIONS

The incidence of AVN did not decline significantly over time in the renal transplant population. Patients with allograft rejection, African American race, peritoneal dialysis and earlier date of transplant were at the highest risk of AVN, while diabetic recipients were at a decreased risk.

Adjusting surgical mortality rates for patient comorbidities: more harm than good?

Background. Studies of medical admissions have questioned the validity of using claims data to adjust for preexisting medical conditions (comorbidities), but the impact of using comorbidities from claims data to risk-adjust mortality rates for high-risk surgery is not well characterized. The purpose of this study was to evaluate the relationship between comorbidities and mortality in administrative data in surgical populations and identify better risk-adjustment methods. Methods. Using the national Medicare database (1994-1997), we identified admissions for elective abdominal aortic aneurysm repair (140,577) and pancreaticoduodenectomy (10,530). We calculated the relative risk of mortality (adjusted for age, sex, race, and admission acuity) for 5 chronic conditions that are known (from clinical series) to increase the risk of postoperative mortality and are commonly used in claims-based risk-adjustment models. To explore the potential value of alternative risk-adjustment strategies, we examined relationships between surgical mortality and comorbidities using diagnosis codes identified from previous admissions. Results. Overall, in-hospital mortality for elective abdominal aortic aneurysm (AAA) repair and pancreaticoduodenectomy were 5.1% and 10.4%, respectively. For both procedures, 3 of the 5 comorbidities were associated with decreased risk of mortality: prior myocardial infarction (MI) [RR = 0.38; 95% confidence interval (CI), 0.33-0.43 for AAA; RR = 0.38; 95% CI, 0.21-0.69 for pancreaticoduodenectomy), malignancy (RR = 0.67; 95% CI, 0.59-0.76 for AAA; RR = 0.74; 95% CI, 0.45-1.21 for pancreaticoduodenectomy], and diabetes (RR = 0.76; 95% CI, 0.64-0.84 for AAA; RR = 0.59; 95% CI, 0.49-0.69 for pancreaticoduodenectomy). Using comorbidities identified from prior admissions increased the mortality risk estimates for prior MI (RR = 1.22; 95% CI, 1.08-1.38 for AAA; RR = 0.80; 95% CI, 0.49-1.30 for pancreaticoduodenectomy) and diabetes (RR = 1.41; 95% CI, 1.25-1.59 for AAA; RR = 0.94; 95% CI, 0.78-1.14 for pancreaticoduodenectomy). Conclusions. Because comorbidities coded on the index admission appear protective, incorporating them in risk-adjustment models for studies comparing surgical performance may penalize providers for taking care of sicker patients. When available, comorbidity information from prior hospitalizations may be more useful for risk adjustment.

Hospitalizations for total hip arthroplasty after renal transplantation in the United States

The national incidence of and factors associated with total hip arthroplasty in renal transplant recipients has not been reported. We conducted an historical cohort study of 42096 renal transplant recipients in the United States between 1 July 1994 and 30 June 1998. Primary outcomes were associations with hospitalizations for a primary discharge code of total hip arthroplasty (ICD9 procedure code 81.51x) within 3 years after renal transplant using Cox regression. Renal transplant recipients had a cumulative incidence of total hip arthroplasty of 5.1 episodes/1000 person-years, which is 5-8 times higher than reported in the general population. Avascular necrosis of the hip was the most frequent primary diagnosis associated with total hip arthroplasty in this population (72% of cases). Repeat surgeries were performed in 27% of patients with avascular necrosis, vs. 15% with other diagnoses. Total hip arthroplasty was more frequent in transplant recipients who were older, African American, or who experienced allograft rejection. Mortality after total hip arthroplasty was 0.21% at 30 days and 15% at 3 years, similar to the mortality of all transplant recipients. The most common indication for total hip arthroplasty after renal transplant is avascular necrosis of the hip, in contrast to the general population. Although repeat surgeries are common, total hip arthroplasty is well tolerated and is not associated with increased mortality in this population.

Adjuvant chemotherapy use for Medicare beneficiaries with stage II colon cancer

PURPOSE

Clinical trials have not demonstrated that adjuvant chemotherapy improves survival for patients with resected stage II colon cancer. Nevertheless, patients may receive this treatment despite its uncertain benefit. The objective of this study was to determine the extent to which adjuvant chemotherapy is used for patients with stage II colon cancer.

PATIENTS AND METHODS

Using the Surveillance, Epidemiology, and End Results-Medicare linked database, we identified 3,151 patients aged 65 to 75 with resected stage II colon cancer and no adverse prognostic features. The primary outcome was chemotherapy use within 3 months of surgery ascertained from claims submitted to Medicare. Relationships between patient characteristics and adjuvant chemotherapy use were measured and their significance was assessed using multivariable logistic regression. Survival for treated and untreated patients was compared using a Cox model.

RESULTS

Twenty-seven percent of patients received chemotherapy during the 3 postoperative months. Younger age at diagnosis, white race, unfavorable tumor grade, and low comorbidity were each associated with a greater likelihood of receiving treatment. Sex, the number of examined lymph nodes in the tumor specimen, the urgency of the surgical admission, and median income was each unrelated to treatment. Five-year survival was 75% for untreated patients and 78% for treated patients. After adjusting for known between-group differences, the hazard ratio for survival associated with adjuvant treatment was 0.91 (95% confidence interval, 0.77 to 1.09).

CONCLUSION

A substantial percentage of Medicare beneficiaries with resected stage II colon cancer receive adjuvant chemotherapy despite its uncertain benefit.

Coding of diagnoses, comorbidities, and complications of total hip arthroplasty

International Classification of Diseases coding of patient charts is used by hospitals to allow for billing of patients. Coding information also is used for assessing physician effectiveness. The purpose of the current study was to examine hospital coding for patients having total hip arthroplasty. One hundred consecutive primary total hip replacements were done at one medical center by two orthopaedic surgeons. Patient charts were coded by hospital coders according to the Health Care Finance Administration guidelines. Subsequently, an orthopaedist-based team did a secondary review of these charts and the two sets of codes were compared. The diagnostic codes were similar between the two groups for 87% (174 of 200 codes) of the cases. Comorbidities generally were undercoded by the hospital coders who reported 2.9 comorbidities per patient, whereas the secondary review reported 3.7 comorbidities per patient. The hospital coders found a complication rate of 1.2 per patient, whereas the secondary review revealed a rate of 0.4 per patient. Based on the results of the current study, the authors conclude that it is important to ensure three issues regarding the standard of coding and quality control: (1) the qualifications of the coders; (2) an interaction between coders and healthcare professionals to check that coding is accurate and reproducible; and (3) communication among various health professionals (including the primary surgeon) and coders to determine what actually are appropriate diagnoses, comorbidities, and complications.

Is coding of diagnoses, comorbidities, and complications in total knee arthroplasty accurate?

Coding of diagnoses, comorbidities, and complications is important for health care delivery, not only for appropriate hospital and physician reimbursement, but also for a correct assessment of complication rates. The purpose of this study was to determine the agreement of coding of diagnoses, comorbidities, and complications for total knee arthroplasty between 2 groups of coders. Between January 1, 1997, and November 18, 1997, 100 consecutive primary total knee arthroplasties were done by 2 orthopaedic surgeons. Diagnoses, comorbidities, and complications were coded by professional hospital coders according to the Healthcare Finance Administration guidelines, then recoded by a second team with orthopaedic experience. Although the hospital coders matched diagnoses with the orthopaedic team 96.5% of the time, they determined a complication rate of 1.4 per patient and a comorbidity rate of 2.9 per patient, whereas the orthopaedic team coded for 0.7 complications per patient and 3.7 comorbidities. Based on these results, there should be interaction and communication between hospital coders and health care professionals to check that coding is accurate and reproducible.

Is surgery getting safer? National trends in operative mortality

BACKGROUND

Although mortality rates for some cardiovascular procedures seem to have declined, it is unclear whether other high-risk procedures are becoming safer over time.

STUDY DESIGN

We examined national trends between 1994 and 1999 in operative mortality for 14 high-risk cardiovascular and cancer procedures in the national population of Medicare beneficiaries over age 65. Secular trends were examined using logistic regression adjusting for age, gender, race, socioeconomic status, admission acuity, comorbidities, and hospital volume.

RESULTS

Observed mortality rates varied widely across the 14 procedures, from 2% (carotid endarterectomy) to 16% (esophagectomy). Over the 6-year study period, average patient age increased for all procedures, and patients were more likely to undergo operation at high-volume hospitals for some procedures (pancreatic resection, esophagectomy, cystectomy, and pneumonectomy). After accounting for these changes, operative mortality declined significantly for three cardiovascular procedures, as evidenced by adjusted odds ratios (OR) for the 6-year effect on operative mortality (coronary artery bypass graft OR = 0.85, 95% confidence interval [CI] 0.81 to 0.88; carotid endarterectomy OR = 0.86,95% CI 0.80 to 0.93; mitral valve replacement OR = 0.89, 95% CI 0.81 to 0.97). In contrast, operative mortality did not decline for any of the cancer procedures. In fact, adjusted mortality increased for colectomy for colon cancer (OR= 1.13, 95% CI 1.07 to 1.19).

CONCLUSIONS

Although risks of some cardiovascular procedures are declining over time, there is no evidence that other types of high-risk surgery are becoming safer. These findings suggest the need for systematic efforts to monitor and improve surgical performance.

Validity of information on comorbidity derived rom ICD-9-CCM administrative data

BACKGROUND

The comorbidity variables that constitute the Charlson index are widely used in health care research using administrative data. However, little is known about the validity of administrative data in these comorbidities. The agreement between administrative hospital discharge data and chart data for the recording of information on comorbidity was evaluated. The predictive ability of comorbidity information in the two data sets for predicting in-hospital mortality was also compared.

METHODS

One thousand two hundred administrative hospital discharge records were randomly selected in the region of Calgary, Alberta, Canada in 1996 and used a published coding algorithm to define the 17 comorbidities that constitute the Charlson index. Corresponding patient charts for the selected records were reviewed as the "criterion standard" against which validity of the administrative data were judged.

RESULTS

Compared with the chart data, administrative data had a lower prevalence in 10 comorbidities, a higher prevalence in 3 and a similar prevalence in 4. The kappa values ranged from a high of 0.87 to a low of 0.34; agreement was therefore near perfect for one variable, substantial for six, moderate for nine, and only fair for one variable. For the Charlson index score ranging from 0 to 5 to 6 or higher, agreement was moderate to substantial (kappa = 0.56, weighted kappa = 0.71). When 16 Charlson comorbidities from administrative data were used to predict in-hospital mortality, 10 comorbidities and the index scores defined using administrative data yielded odds ratios that were similar to those derived from chart data. The remaining six comorbidities yielded odds ratios that were quite different from those derived from chart data.

CONCLUSIONS

Administrative data generally agree with patient chart data for recording of comorbidities although comorbidities tend to be under-reported in administrative data. The ability to predict in-hospital mortality is less reliable for some of the individual comorbidities than it is for the summarized Charlson index scores in administrative data.

Epidemiology of foot and ankle injuries in the workplace

Injuries to the ankle, foot, and toes of injured workers are common. Although the workers' compensation system attempts to collect information about these injuries and report them in a meaningful way, the process is somewhat inconsistent, inaccurate, and provides information relevant to government agencies but not to physician organizations seeking to improve the prevention and treatment of work-related injuries. Several improvements in data collection could greatly increase the system's efficiency and usefulness. The first is to establish a national system of uniform data collection from each State. The second improvement involves standardization of the forms and means that the data are collected from the worker including a system to retrieve information that has been omitted during the initial encounter. The third, and most important, refinement is to change the reporting of the illness to match currently accepted medical diagnosis codes (ICD-9). Using the current system and making these improvements, the United States would be able to collect more meaningful data on work injuries in this country. From that point forward medical interventions could then be created and their effects more meaningfully analyzed. All those involved in the care and treatment of the injured worker must recognize not only the limitations of the collected data but also how this information can be manipulated. For example, whereas the total cost of benefits paid by employers increased during one decade the average cost per covered employee decreased. Employers and insurers can argue that their total costs have consistently risen but only by knowing that the cost per covered employee had decreased during that same time period can that argument be countered. Similarly, implementation of certain safety practices during a given time period may be accompanied by a reduction in the number of injuries. Changes in the laws that altered injury reporting and reduced the coverage for certain injuries, however, may have created a situation where the reported injury rates were reduced by accounting practice, not medical practice. It is also well established that insurance company profits may be linked as much to the general economy and stock market as they are to premiums and claims. It is hoped that this article begins an understanding of the extent of the problem of foot and ankle injuries in the working population. Additional statistical information on specific topics is presented elsewhere in this issue.

Using administrative data to study persons with disabilities

Administrative data result from administering health plans--tracking service utilization, paying claims, monitoring costs and quality--and have been used extensively for health services research. This article examines the strengths and limitations of administrative data for health services research studies of people with disabilities. Administrative data offer important advantages: encompassing large populations over time, ready availability, low cost, and computer readability. Questions arise about how to identify people with disabilities, capture disability-related services, and determine meaningful health care outcomes. Potentially useful administrative data elements include eligibility for Medicare or Medicaid through Social Security disability determinations, diagnosis and procedure codes, pharmacy claims, and durable medical equipment claims. Linking administrative data to survey or other data sources enhances the utility of administrative data for disability studies.

Lessons learnt about early prostate cancer from large scale databases: population-based pearls of wisdom

Prostate cancer is one of most common solid tumors in men and poses some of the most difficult problems in clinical research. Although many clinical research hypotheses in this condition have been explored using single center cases series and multi-center clinical trials, the results of these studies have often been equivocal, leaving many questions unanswered. Recently, investigators have utilized large, administrative datasets for prostate cancer research. These databases tend to include large numbers of patients from different geographic regions increasing their generalizability and statistical power. The goal of this report is to review lessons learnt about early prostate cancer using these data sources. In particular, we focus on the application of large, population-based datasets to address issues concerning the natural history of prostate cancer, the impact of race on outcomes in prostate cancer and the effectiveness of various treatments for localized disease. Information gathered from large, administrative databases will be helpful when counseling patients regarding their treatments options for localized prostate cancer and in identifying future directions for prostate cancer research.

Chronic dialysis patients have high risk for pulmonary embolism

Pulmonary embolism has been considered uncommon in chronic dialysis patients, but has not been adequately studied in a large population. In the US Renal Data System (USRDS), 76,718 patients presenting with end-stage renal disease (ESRD) between January 1, 1996, and December 31, 1996, were analyzed in an historical cohort study. The outcome was hospitalizations with a primary discharge diagnosis of pulmonary embolism (International Classification of Diseases, Ninth Revision code 415.1x) occurring within 1 year of the first ESRD treatment and excluding those occurring after renal transplantation. For dialysis patients, hospitalization rates for pulmonary embolism were obtained from the hospitalization section of the 1999 USRDS. For the general population, hospitalization rates for pulmonary embolism were obtained from the National Hospital Discharge Survey for 1996. Comorbidities from the Medical Evidence Form (Centers for Medicare and Medicaid Services, previously known as the Health Care Financing Administration; form 2728) were used to generate approximated stratified models of adjusted incidence ratios for pulmonary embolism (comorbidities could not be stratified for the general population). In 1996, the overall incidence rate of pulmonary embolism was 149.90/100,000 dialysis patients compared with 24.62/100,000 persons in the US population, with an age-adjusted incidence ratio of 2.34 in dialysis patients. Younger dialysis patients had the greatest relative risk for pulmonary embolism. The age-adjusted incidence ratio of pulmonary embolism after excluding dialysis patients with known risk factors for pulmonary embolism was 2.11. Ninety-five percent confidence intervals for all age categories in both models were statistically significant. Chronic dialysis patients have high risk for pulmonary embolism, independent of comorbidity.

Hospital volume and surgical mortality in the United States

BACKGROUND

Although numerous studies suggest that there is an inverse relation between hospital volume of surgical procedures and surgical mortality, the relative importance of hospital volume in various surgical procedures is disputed.

METHODS

Using information from the national Medicare claims data base and the Nationwide Inpatient Sample, we examined the mortality associated with six different types of cardiovascular procedures and eight types of major cancer resections between 1994 and 1999 (total number of procedures, 2.5 million). Regression techniques were used to describe relations between hospital volume (total number of procedures performed per year) and mortality (in-hospital or within 30 days), with adjustment for characteristics of the patients.

RESULTS

Mortality decreased as volume increased for all 14 types of procedures, but the relative importance of volume varied markedly according to the type of procedure. Absolute differences in adjusted mortality rates between very-low-volume hospitals and very-high-volume hospitals ranged from over 12 percent (for pancreatic resection, 16.3 percent vs. 3.8 percent) to only 0.2 percent (for carotid endarterectomy, 1.7 percent vs. 1.5 percent). The absolute differences in adjusted mortality rates between very-low-volume hospitals and very-high-volume hospitals were greater than 5 percent for esophagectomy and pneumonectomy, 2 to 5 percent for gastrectomy, cystectomy, repair of a nonruptured abdominal aneurysm, and replacement of an aortic or mitral valve, and less than 2 percent for coronary-artery bypass grafting, lower-extremity bypass, colectomy, lobectomy, and nephrectomy.

CONCLUSIONS

In the absence of other information about the quality of surgery at the hospitals near them, Medicare patients undergoing selected cardiovascular or cancer procedures can significantly reduce their risk of operative death by selecting a high-volume hospital.

Identifying patient preoperative risk factors and postoperative adverse events in administrative databases: results from the Department of Veterans Affairs National Surgical Quality Improvement Program

BACKGROUND

The Department of Veterans Affairs (DVA) National Surgical Quality Improvement Program (NSQIP) employs trained nurse data collectors to prospectively gather preoperative patient characteristics and 30-day postoperative outcomes for most major operations in 123 DVA hospitals to provide risk-adjusted outcomes to centers as quality indicators. It has been suggested that routine hospital discharge abstracts contain the same information and would provide accurate and complete data at much lower cost.

STUDY DESIGN

With preoperative risks and 30-day outcomes recorded by trained data collectors as criteria standards, ICD-9-CM hospital discharge diagnosis codes in the Patient Treatment File (PTF) were tested for sensitivity and positive predictive value. ICD-9-CM codes for 61 preoperative patient characteristics and 21 postoperative adverse events were identified.

RESULTS

Moderately good ICD-9-CM matches of descriptions were found for 37 NSQIP preoperative patient characteristics (61%); good data were available from other automated sources for another 15 (25%). ICD-9-CM coding was available for only 13 (45%) of the top 29 predictor variables. In only three (23%) was sensitivity and in only four (31%) was positive predictive value greater than 0.500. There were ICD-9-CM matches for all 21 NSQIP postoperative adverse events; multiple matches were appropriate for most. Postoperative occurrence was implied in only 41%; same breadth of clinical description in only 23%. In only four (7%) was sensitivity and only two (4%) was positive predictive value greater than 0.500.

CONCLUSION

Sensitivity and positive predictive value of administrative data in comparison to NSQIP data were poor. We cannot recommend substitution of administrative data for NSQIP data methods.

Retrospective hospital-based searches for cases of acute flaccid paralysis

OBJECTIVE

Australia had to demonstrate adequate acute flaccid paralysis (AFP) surveillance by achieving a rate of one per 100,000 in children under the age of 15 to fulfil one of the requirements of the Regional Commission for the Certification of Poliomyelitis Eradication to be declared polio free. To increase the ascertainment rate of AFP cases, a hospital search was conducted to identify cases not reported to the active AFP surveillance.

METHODS

A computerised search of hospital admissions in New South Wales (NSW) and Western Australia (WA) on ICD-9 codes of Guillain Barré Syndrome (GBS), unspecified encephalitis, poliomyelitis, vaccine-associated paralytic polio (VAPP) and flaccid paralysis was conducted for the period 1995-98. Medical records of cases that were not reported to the active surveillance were reviewed in three hospitals of NSW and two hospitals in WA.

RESULTS

Twenty additional cases recorded as GBS and five as transverse myelitis (TM) were identified through the searches, which increased the average four-year AFP rate from 1.0 to 1.4 per 100,000 in children under the age of 15 years in these two states and the overall AFP rate in Australia increased from 0.78 to 1.14. There were no cases of polio or VAPP found. Nine cases of GBS and five of TM reported to the active AFP surveillance were not found in the hospital searches.

CONCLUSION

A combination of active surveillance and hospital-based searches increased the investigated AFP rate, which fulfilled one of the certification requirements for Australia to be certified polio free.

IMPLICATIONS

Until global certification is achieved, AFP surveillance needs to be improved to identify cases of importation of wild poliovirus.

Who gets adjuvant treatment for stage II and III rectal cancer? Insight from surveillance, epidemiology, and end results--Medicare

PURPOSE

To examine the relationship between patient characteristics and the use of adjuvant pelvic radiation with and without chemotherapy among patients aged 65 years and older with stage II and III rectal cancer.

PATIENTS AND METHODS

A retrospective cohort study using the Surveillance, Epidemiology, and End Results-Medicare linked database identified 1,411 patients aged 65 and older with resected stage II and III rectal cancers diagnosed between 1992 and 1996. From claims submitted to Medicare, we measured the use of pelvic radiation therapy with or without chemotherapy and pre- or postoperatively.

RESULTS

Fifty-seven percent of patients received radiation, 42% received chemotherapy and radiation, and 7% had treatment delivered preoperatively. Age was the strongest determinant of treatment: 73% of patients aged 65 to 69, 66% aged 70 to 75, 52% aged 75 to 79, 39% aged 80 to 84, and 21% aged 85 to 89 received radiation. The age trend remained strong after adjusting for other factors that predict receipt of treatment and after exclusion of patients with any evident comorbidity (P <.001). Patients were more likely to receive radiation treatment if they had an abdominal perineal resection, stage III disease, or a T4 tumor.

CONCLUSION

Because pelvic recurrences are a substantial cause of morbidity, further efforts are needed to ensure that elderly patients have the opportunity to make informed decisions regarding adjuvant treatment.

Quality of the information contained in the minimum basic data set: results from an evaluation in eight hospitals

To assess the quality of the information included in the minimum basic data set (MBDS) of the eight public hospitals of the Murcia region in order to ascertain what should be improved to be valid and reliable. An external encoder performed a recoding of a random sample of hospital discharges, using the patients hospital records and comparing afterwards the information obtained with the one reflected in the MBDS databases. Quality was assessed using 12 criteria. The reviewed discharges sample consisted at least of 96 cases per hospital (Type I error = 0.05, Type II = 0.10, for the most unfavorable case). A total of 796 cases were reviewed. The MBDS disagreement percentages with the patient record data were higher for the clinical data, with 41.6% for the main diagnosis and 33.5% for the main surgical procedure, being in both cases higher in those hospitals that had used to codify just the discharge record with regard to those that did so with the complete patient record. The variation rate in the diagnosis-related group (DRG) assignment was of 29.6%, and there was a decrease in the case-mix index of 1.07397 when reviewing with the patient record to 1.05555 in the MBDS. Within the administrative data, the highest disagreement rate was for the physician that signs the discharge (60.5%) and the patient's address (31.6%). In many of these assessed aspects there are significant differences between hospitals. A reliability problem was identified in the collected data, which mainly affects the clinical variables. It is therefore advisable to carefully assess the use of this information (both the MBDS directly as well as its grouping through the use of patient classification systems), and the indicators derived from it as its quality is not guaranteed. Systematic assessment and quality control of the MBDS production is advised.

Possible effect of DRGs on the classification of stroke: implications for epidemiological surveillance

BACKGROUND AND PURPOSE

Accurate data on the distribution of stroke subtypes are essential for understanding the forces driving recent morbidity and mortality trends. The introduction of diagnosis-related groups (DRGs) in the 1980s may have affected the distribution of stroke subtypes as defined by International Classification of Diseases, Ninth Revision (ICD-9), discharge diagnosis codes.

METHODS

The Pawtucket Heart Health Program cardiovascular surveillance data were used to examine trends in stroke classification for 1980 to 1991 in relation to the introduction of DRGs in 2 communities in Massachusetts and Rhode Island, where DRGs were implemented 2 years apart. Included were all hospital discharges for residents aged 35 to 74 with a primary ICD-9 diagnosis of 431 to 432, 434, or 436 to 437 (N=1386 in Rhode Island, N=1839 in Massachusetts).

RESULTS

In each state, concurrently with the introduction of DRGs, the proportion of strokes classified as cerebral occlusion (ICD-9 434.0 to 434.9) increased, and the proportion classified as acute but ill-defined (ICD-9 436.0 to 436.9) decreased. Before DRGs, 30.0% of strokes in Rhode Island and 26.6% in Massachusetts were classified as cerebral occlusion, whereas 51.8% in Rhode Island and 51.7% in Massachusetts were classified as acute ill defined. After DRGs were instituted, the proportions of cerebral occlusion and acute, ill-defined stroke, respectively, were 70.9% and 8.5% in Rhode Island and 74.1% and 7.7% in Massachusetts (chi(2), all P<0.001). The proportions of strokes classified as intracerebral hemorrhage or transient cerebral ischemia remained constant.

CONCLUSIONS

The implementation of DRGs may have influenced coding of strokes to the ICD-9. Findings highlight the limitations of hospital discharge data for evaluating stroke subtypes and demonstrate the need for community-based surveillance for monitoring specific trends in stroke.

Age and adjuvant chemotherapy use after surgery for stage III colon cancer

BACKGROUND

Randomized trials have established that 5-fluorouracil-based adjuvant chemotherapy following resection of stage III colon cancer reduces subsequent mortality by as much as 30%. However, the extent to which adjuvant therapy is used outside the clinical trial setting, particularly among the elderly, is unknown.

METHODS

A retrospective cohort study utilizing the Surveillance, Epidemiology, and End Results/Medicare-linked database identified 6262 patients aged 65 years and older with resected stage III colon cancer. The primary outcome was chemotherapy use within 3 months of surgery, as ascertained from Medicare claims. We examined the extent to which age at diagnosis was associated with adjuvant chemotherapy usage, and we adjusted for potential confounding based on differences in other patient characteristics with the use of multiple logistic regression. All P values were two-sided.

RESULTS

Age at diagnosis was the strongest determinant of chemotherapy: 78% of patients aged 65-69 years, 74% of those aged 70-74 years, 58% of those aged 75-79 years, 34% of those aged 80-84 years, and 11% of those aged 85-89 years received postoperative chemotherapy. The age trend remained pronounced after adjustment for potential confounding based on variation in patients' demographic and clinical characteristics and after exclusion of patients with any evident comorbidity (all P values <.001).

CONCLUSIONS

Adjuvant chemotherapy for stage III colon cancer is used extensively, especially for patients under the age of 75 years. However, treatment rates decline dramatically with chronologic age. Because patients in their 70s and even 80s have a reasonable life expectancy, further efforts are needed to ensure that elderly patients have the opportunity to make informed decisions regarding this potentially curative treatment.

Validating billing data for RBC transfusions: a brief report

BACKGROUND

Administrative data are used often for research, but without validation of their accuracy. The validity of the billing for blood transfusion was assessed in one tertiary-care hospital.

MATERIALS AND METHODS

Patient discharge data were retrieved from a database containing demographics, diagnoses, and charges. There was random selection of 358 patients who were billed for RBC transfusion and 358 who were not, within a 2-month period. The blood bank's transfusion records were reviewed. Sensitivity was defined as the proportion of transfused patients who were billed, and specificity as the proportion of nontransfused patients who were not billed. Patient characteristics were compared by using Wilcoxon's rank sum test and the chi-square test.

RESULTS

Sixty-one transfused patients were not billed for the transfusion. No patient was billed without transfusion. Thus, the sensitivity and specificity were 83 percent (95% CI, 79-87%) and 100 percent, respectively. Nine patients who were not issued RBCs were appropriately not billed for RBCs, although the billing record suggests they had a procedure involving transfusion. These patients were called true-negative. The patients not billed were older (58 years vs. 55 years; p = 0.046) and less likely to have commercial insurance (5% vs. 15%; p = 0.035) than billed patients.

CONCLUSIONS

The billing for RBC transfusion in one large institution is reassuringly valid. The specificity is excellent, and the sensitivity is higher than that seen in other studies of coding validity.

Introducing DRG-based financing in Hungary: a study into the relationship between supply of hospital beds and use of these beds under changing institutional circumstances

Most hospital reforms carried out in Europe over the past few decades concern the supply of hospital beds and hospital financing systems. In Hungary, financing was not tied to hospital input or output until a Diagnosis-Related-Group system was introduced. This change provided an opportunity to study the effect of the new system, taking the supply of hospital beds into account. We studied the effect of the financing system and bed supply on four output parameters, average length of stay; admission rate; occupancy; and case-mix. The incentives of the financing system influenced the length of stay (shorter) and the admission rate (more admissions). Although the case-mix did increase, occupancy was not affected. The supply of more beds resulted in higher admission rates and a slightly lower efficiency (a lower occupancy rate). No interaction effects of (variations in) the bed supply and the financing system were found.

Development of a comorbidity index using physician claims data

Important comorbidities recorded on outpatient claims in administrative datasets may be missed in analyses when only inpatient care is considered. Using the comorbid conditions identified by Charlson and colleagues, we developed a comorbidity index that incorporates the diagnostic and procedure data contained in Medicare physician (Part B) claims. In the national cohorts of elderly prostate (n = 28,868) and breast cancer (n = 14,943) patients assessed in this study, less than 10% of patients had comorbid conditions identified when only Medicare hospital (Part A) claims were examined. By incorporating physician claims, the proportion of patients with comorbid conditions increased to 25%. The new physician claims comorbidity index significantly contributes to models of 2-year noncancer mortality and treatment received in both patient cohorts. We demonstrate the utility of a disease-specific index using an alternative method of construction employing study-specific weights. The physician claims index can be used in conjunction with a comorbidity index derived from inpatient hospital claims, or employed as a stand-alone measure.

Continuous training as a key to increase the accuracy of administrative data

The aim of this study was to evaluate the impact of a program of training, education and awareness on the accuracy of the data collected from hospital discharge abstracts. Four random samples of hospital discharge abstracts relating to four different periods were studied. The evaluation of the impact of systematic training and education activities was performed by checking the quality of abstracting information from the medical records. The analysis was carried out at the Istituto Dermopatico dell'Immacolata, a research hospital (335 beds) in Rome, Italy, which specializes in dermatology, plastic and vascular surgery. Error rates in discharge abstracts were subdivided into six categories: selection of the wrong principal diagnosis (type A); low specificity of the principal diagnosis (type B); incomplete reporting of secondary diagnoses (type C); selection of the wrong principal procedure (type D); low specificity of the principal procedure (type E); incomplete reporting of procedures (type F). A specific rate for errors modifying classification in diagnosis related groups (DRG) was then estimated and the effect of re-abstracting on the case-mix index evaluated. Error types A, B, C, E and F dropped from 8.5% to 2%, 15.8 to 4.9, 31.8 to 13.1,4.1 to 0.3 and 22 to 2.6%, respectively. Error type D was 0.7 both in the first (the baseline) and fourth periods of analysis. All differences in error types were statistically significant. In 1999 8.3% of cases were assigned to a different DRG after re-abstracting as compared with 24.3% in the third quarter of 1994, 23.8% in the first quarter of 1995 and 5.5% in September-October 1997. Continuous training and feedback of information to departments have shown to be successful in improving the quality of abstracting information at patient level from the medical record. These positive results were facilitated by the introduction of a prospective payment system to finance inpatient hospital activity. The effort to increase administrative data quality at hospital level facilitates the use of those data sets for internal quality management activities.

Socioeconomic issues affecting the treatment of obesity in the new millennium

The prevalence of obesity among the populations of most developed countries has increased to such an extent that the healthcare and social security/disability system will accumulate direct and indirect costs related to obesity that will be more substantial than those for any other primary disease within this generation. For the past decade, the Healthcare Financing Agency, which oversees the Medicare and Medicaid programmes, has required all physicians and healthcare agencies serving beneficiaries of these programmes to include diagnoses using codes established by the ninth revision of the World Health Organization's International Classification of Diseases. This coding system actually distorts data collection and undermines appropriate medical insurance reimbursement for the treatment of obesity. Societal prejudices, inability of governmental agencies to address future concerns and the business community's attempts to control healthcare costs without addressing the underlying issues contributing to these costs have led to confusion on how to confront this emerging epidemic. How will we develop the scientific knowledge and the political willpower to confront this epidemic? First, we need more accurate methods for classifying obesity and for measuring the cost of treatment. We can then determine if it is more cost effective to prevent or treat obesity early in its evolution or pay for its consequences in the form of treatment costs associated with its multiple comorbid diseases, such as hypertension, other cardiovascular disorders, diabetes mellitus, osteoarthritis and cancers, plus the lost productivity from absenteeism, premature retirement and death.

Automated computerized intensive care unit severity of illness measure in the Department of Veterans Affairs: preliminary results. SISVistA Investigators. Scrutiny of ICU Severity Veterans Health Sysyems Technology Architecture

OBJECTIVE

To evaluate the feasibility of an automated intensive care unit (ICU) risk adjustment tool (acronym: SISVistA) developed by selecting a subset of predictor variables from the Acute Physiology and Chronic Health Evaluation (APACHE) III available in the existing computerized database of the Department of Veterans Affairs (VA) healthcare system and modifying the APACHE diagnostic and comorbidity approach.

DESIGN

Retrospective cohort study.

SETTING

Six ICUs in three Ohio Veterans Affairs hospitals.

PATIENT SELECTION

The first ICU admission of all patients from February 1996 through July 1997.

OUTCOME MEASURE

Mortality at hospital discharge.

METHODS

The predictor variables, including age, comorbidity, diagnosis, admission source (direct or transfer), and laboratory results (from the +/- 24-hr period surrounding admission), were extracted from computerized VA databases, and APACHE III weights were applied using customized software. The weights of all laboratory variables were added and treated as a single variable in the model. A logistic regression model was fitted to predict the outcome and the model was validated using a boot-strapping technique (1,000 repetitions).

MAIN RESULTS

The analysis included all 4,651 eligible cases (442 deaths). The cohort was predominantly male (97.5%) and elderly (63.6 +/- 12.0 yrs). In multivariate analysis, significant predictors of hospital mortality included age (odds ratio [OR], 1.06; 95% confidence interval [CI], 1.04-1.09), comorbidity (OR, 1.11; 95% CI, 1.08-1.15), total laboratory score (OR, 1.07; 95% CI, 1.06-1.08), direct ICU admission (OR, 0.39; 95% CI, 0.31-0.49), and several broad ICU diagnostic categories. The SISVistA model had excellent discrimination and calibration (C statistic = 0.86, goodness-of-fit statistics; p > .20). The area under the receiver operating characteristic curve of the validated model was 0.86.

CONCLUSIONS

Using common data elements often found in hospital computer systems, SISVistA predicts hospital mortality among patients in Ohio VA ICUs. This preliminary study supports the development of an automated ICU risk prediction system on a more diverse population.

Does clinical evidence support ICD-9-CM diagnosis coding of complications?

BACKGROUND

Hospital discharge diagnoses, coded by use of the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM), increasingly determine reimbursement and support quality monitoring. Prior studies of coding validity have investigated whether coding guidelines were met, not whether the clinical condition was actually present.

OBJECTIVE

To determine whether clinical evidence in medical records confirms selected ICD-9-CM discharge diagnoses coded by hospitals.

RESEARCH DESIGN AND SUBJECTS

Retrospective record review of 485 randomly sampled 1994 hospitalizations of elderly Medicare beneficiaries in Califomia and Connecticut.

MAIN OUTCOME MEASURE

Proportion of patients with specified ICD-9-CM codes representing potential complications who had clinical evidence confirming the coded condition.

RESULTS

Clinical evidence supported most postoperative acute myocardial infarction diagnoses, but fewer than 60% of other diagnoses had confirmatory clinical evidence by explicit clinical criteria; 30% of medical and 19% of surgical patients lacked objective confirmatory evidence in the medical record. Across 11 surgical and 2 medical complications, objective clinical criteria or physicians' notes supported the coded diagnosis in >90% of patients for 2 complications, 80% to 90% of patients for 4 complications, 70% to <80% of patients for 5 complications, and <70% for 2 complications. For some complications (postoperative pneumonia, aspiration pneumonia, and hemorrhage or hematoma), a large fraction of patients had only a physician's note reporting the complication.

CONCLUSIONS

Our findings raise questions about whether the clinical conditions represented by ICD-9-CM codes used by the Complications Screening Program were in fact always present. These findings highlight concerns about the clinical validity of using ICD-9-CM codes for quality monitoring.

An introduction to epidemiological research with medical databases

In most regards, database research is like any other epidemiological endeavor: excellent research can be conducted, but there are many potential difficulties. Training in appropriate epidemiological and statistical methodology, together with knowledge of the databases and their coding systems, is likely to magnify the advantages of databases and also minimize the potential problems. As in all epidemiological investigations, the quality of the data and the methodology employed need to be carefully considered in the context of the research questions at hand.

Drinking water turbidity and gastrointestinal illness in the elderly of Philadelphia

STUDY OBJECTIVE

To investigate the association between drinking water quality and gastrointestinal illness in the elderly of Philadelphia.

DESIGN

Within the general population, children and the elderly are at highest risk for gastrointestinal disease. This study investigates the potential association between daily fluctuations in drinking water turbidity and subsequent hospital admissions for gastrointestinal illness of elderly persons, controlling for time trends, seasonal patterns, and temperature using Poisson regression analysis.

SETTING AND PARTICIPANTS

All residents of Philadelphia aged 65 and older in 1992-1993 were studied through their MEDICARE records.

MAIN RESULTS

For Philadelphia's population aged 65 and older, we found water quality 9 to 11 days before the visit was associated with hospital admissions for gastrointestinal illness, with an interquartile range increase in turbidity being associated with a 9% increase (95% CI 5.3%, 12.7%). In the Belmont service area, there was also an association evident at a lag of 4 to 6 days (9.1% increase, 95% CI 5.2, 13.3). Both associations were stronger in those over 75 than in the population aged 65-74. This association occurred in a filtered water supply in compliance with US standards.

CONCLUSIONS

Elderly residents of Philadelphia remain at risk of waterborne gastrointestinal illness under current water treatment practices. Hospitalisations represent a very small percentage of total morbidity.

Rates of initial and recurrent thromboembolic disease among patients with malignancy versus those without malignancy. Risk analysis using Medicare claims data

Although the association between malignancy and thromboembolic disease is well established, the relative risk of developing initial and recurrent deep vein thrombosis (DVT) or pulmonary embolism (PE) among patients with malignancy versus those without malignancy has not been clearly defined. The Medicare Provider Analysis and Review Record (MEDPAR) database was used for this analysis. Patients hospitalized during 1988-1990 with DVT/PE alone, DVT/PE and malignancy, malignancy alone, or 1 of several nonmalignant diseases (other than DVT/PE) were studied. The association of malignancy and nonmalignant disease with an initial episode of DVT/PE, recurrent DVT/PE, and mortality were analyzed. The percentage of patients with DVT/PE at the initial hospitalization was higher for those with malignancy compared with those with nonmalignant disease (0.6% versus 0.57%, p = 0.001). The probability of readmission within 183 days of initial hospitalization with recurrent thromboembolic disease was 0.22 for patients with prior DVT/PE and malignancy compared with 0.065 for patients with prior DVT/PE and no malignancy (p = 0.001). Among those patients with DVT/PE and malignant disease, the probability of death within 183 days of initial hospitalization was 0.94 versus 0.29 among those with DVT/PE and no malignancy (p = 0.001). The relative risk of DVT/PE among patients with specific types of malignancy is described. This study demonstrates that patients with concurrent DVT/PE and malignancy have a more than threefold higher risk of recurrent thromboembolic disease and death (from and cause) than patients with DVT/PE without malignancy. An alternative management strategy may be indicated for such patients.

Frailty bias in comorbidity risk adjustments of community-dwelling elderly populations

Our objective was to describe distortion in outcome studies due to "frailty bias" or differential susceptibility to adverse health outcomes due to frailties but attributed to other factors. We linked an administrative database to survey data (n = 5934) containing functional, condition, and outcome measures. The disease classification scheme of an empirically derived mortality model was used to categorize 7500 ICD-9-CM codes into five risk levels. Cox and logistic regressions were used to compare outcomes. Commonly employed measures differ in their sensitivity to detect and control frailty bias across a spectrum of major chronic diseases. Survival is inversely related to increasing functional impairment after adjusting for age, race, gender, education, number of comorbid conditions, and highest disease risk occurring during follow-up. Functional status appears to be a superior and essential element for control of the frailty bias that threatens comparability of outcome measures across community-dwelling populations containing chronically-ill disabled elderly.

Misclassification and under-reporting of acute myocardial infarction by elderly persons: implications for community-based observational studies and clinical trials

We investigated the accuracy of self-report of hospitalization for acute myocardial infarction (MI) by elderly persons in a community-based prospective study. Among 3809 persons aged 65 years or older followed up for 6 years, self-reported hospitalization for MI was validated by review of primary records and Medicare diagnoses. Among 147 who self-reported MI and for whom hospital records were available, the diagnosis was confirmed in 79 (54%). Myocardial infarction was not a reason for hospitalization among the remaining 68 participants; misclassification with other cardiovascular diagnoses was common. Medicare diagnosis correlated well with primary hospital records. Using Medicare diagnoses as the standard, the diagnosis of MI was confirmed in 53% of self-reports; the sensitivity and specificity of self-report were 51% and 98%, respectively. False-negative reporting was common because only half of hospitalizations for MI were reported. Self-report of hospitalization for MI by elderly persons in the community may be unreliable for ascertaining trends in cardiovascular diseases.

Can cardiovascular clinical characteristics be identified and outcome models be developed from an in-patient claims database?

The objective of this study was to assess whether administrative (claims) databases can be used to assess clinical variables and predict outcome. Although administrative databases are useful for assessing resource utilization, their utility for assessing clinical information is less certain. Prospectively gathered clinical databases, however, are expensive and not widely available. The UB92 formulation of the hospital bill was used as an administrative source of data and compared with the clinical cardiovascular database at Emory University. The claims database was compared with the clinical database for 11 variables. Outcome models were developed with multivariate methods. A total of 11,883 patients who underwent catheterization (5,255 underwent percutaneous transluminal coronary angioplasty [PTCA] and 3,794 underwent coronary artery bypass surgery [CABG]) between 1991 and 1995 were included. For some variables, the claims database correlated well (diabetes, sensitivity 87%, specificity 99%), whereas for others the claims database was less accurate (peripheral vascular disease, sensitivity 20%, specificity 99%). Uncertain coding in the claims database, which can result in the same code being used for co-morbid states and severity of disease, as well as complications, limited the ability of claims to predict outcome. Clinical databases may also be limited by lack of objectivity and missing data. The utility of claims databases to assess severity of disease and co-morbid states is limited, and outcome modeling and risk assessment from claims databases may be inappropriate and spurious. Developing better data standards and less expensive methods for acquisition of clinical data is necessary for improved outcome assessment.

Predictors of Medicare costs in elderly beneficiaries with breast, colorectal, lung, or prostate cancer

BACKGROUND

Determining the apportionment of costs of cancer care and identifying factors that predict costs are important for planning ethical resource allocation for cancer care, especially in markets where managed care has grown.

DESIGN

This study linked tumor registry data with Medicare administrative claims to determine the costs of care for breast, colorectal, lung and prostate cancers during the initial year subsequent to diagnosis, and to develop models to identify factors predicting costs.

SUBJECTS

Patients with a diagnosis of breast (n = 1,952), colorectal (n = 2,563), lung (n = 3,331) or prostate cancer (n = 3,179) diagnosed from 1985 through 1988.

RESULTS

The average costs during the initial treatment period were $12,141 (s.d. = $10,434) for breast cancer, $24,910 (s.d. = $14,870) for colorectal cancer, $21,351 (s.d. = $14,813) for lung cancer, and $14,361 (s.d. = $11,216) for prostate cancer. Using least squares regression analysis, factors significantly associated with cost included comorbidity, hospital length of stay, type of therapy, and ZIP level income for all four cancer sites. Access to health care resources was variably associated with costs of care. Total R2 ranged from 38% (prostate) to 49% (breast). The prediction error for the regression models ranged from < 1% to 4%, by cancer site.

CONCLUSIONS

Linking administrative claims with state tumor registry data can accurately predict costs of cancer care during the first year subsequent to diagnosis for cancer patients. Regression models using both data sources may be useful to health plans and providers and in determining appropriate prospective reimbursement for cancer, particularly with increasing HMO penetration and decreased ability to capture complete and accurate utilization and cost data on this population.

Electronic medical record in the intensive care unit

The EMR in the ICU has the utility of providing the necessary information to make sound clinical decisions for critically ill patients. For it to be optimized, the EMR must be more than just what is being replicated in the written record or merely a documentation tool; it must add value that supports and enhances clinical decision support. The EMR is too expensive a tool just to be a computer designed to ease documentation and retrieve data faster. Gardner and Huff have suggested that the EMR must answer three questions: Why, What, and So What. The "Why" is relatively easy to answer, but the "What" data to use so that the information is meaningful to a provider and the "So What" are more difficult to answer. Provided one can qualitatively assess "What" information is important for a health care provider, then "So What" becomes an important objective in the empirical quantification of the benefits that the EMR provides. It is clear that to analyze some of the outcomes that health care delivery provides, one needs some mechanism to automate the information at the point of care, particularly now that the regulatory agencies are requiring it. Given the fact that there is no single integrated computerized patient record, this becomes the daunting task for the next century. Making it easier for health care providers to interact with the system and providing them with instantaneous feedback that changes their medical decision so they can deliver better care (clinical pathways, clinical practice guidelines) will be the task required of the next generation of CISs.

Information on radiation treatment in patients with breast cancer: the advantages of the linked medicare and SEER data. Surveillance, Epidemiology and End Results

Several studies have found underutilization of radiotherapy in patients with breast cancer; but there are concerns about the completeness of various databases on radiotherapy. We used the linked Medicare-SEER (Surveillance, Epidemiology and End Results) database to compare information on receipt of radiotherapy after diagnosis of breast cancer. More than 18% of women identified by Medicare data as receiving radiotherapy were not so identified by SEER, and 7% of those identified as receiving radiotherapy by SEER were not identified by Medicare. Risk of discordance on radiotherapy information between the two data sets was especially high in women receiving breast-conserving surgery. The combined SEER-Medicare database gives a more complete picture on the use of radiotherapy. The previously reported geographic variations in the use of radiotherapy for breast cancer may be due in part to underreporting of radiotherapy in some areas.

Screening inpatient quality using post-discharge events

BACKGROUND

Decreasing hospital lengths of stay (LOS) hamper efforts to detect and to definitively treat complications of care. Patients leave before some complications are identified.

OBJECTIVES

To develop a computerized method to screen for hospital complications using readily available administrative data from outpatient and nonacute care within 90 days of discharge.

DESIGN

We developed the Complications Screening Program for Outpatient data (CSP-O) by using diagnosis and procedure codes from Medicare Part A and B claims to define 50 complication screens. Seventeen apply to specific procedural cases, and 33 apply to all adult, acute, medical, or surgical hospitalizations. The CSP-O algorithm examined outpatient, physician office, home health agency, and hospice claims within 90 days following discharge.

SUBJECTS

Seven hundred thirty nine thousand, two hundred and forty eight discharges of Medicare beneficiaries (age range, > or = 65 years) were admitted to 515 hospitals nationwide in 1994.

RESULTS

Complete 90-day, post-discharge windows were present for 62.8% of all and 68.5% of procedural cases. The 33 general screens flagged 13.6% of all cases; only 1.8% of procedural cases were flagged by the 17 procedural screens. When we allowed the CSP-O algorithm to scan information from acute hospital readmissions, flag rates rose to 32.8% for general and 8.7% for procedural complications. Controlling for patient and hospital characteristics, flag rates were considerably higher among the very old and at small and for-profit institutions.

CONCLUSIONS

Whereas several CSP-O findings have construct validity, limitations of claims raise concerns. Regardless of the CSPO's ultimate utility, examining post-discharge experiences to identify inpatient complications remains important as LOSs fall.

The use of automated data to identify complications and comorbidities of diabetes: a validation study

We evaluated the accuracy of administrative data for identifying complications and comorbidities of diabetes using International Classification of Diseases, 9th edition, Clinical Modification and Current Procedural Terminology codes. The records of 471 randomly selected diabetic patients were reviewed for complications from January 1, 1993 to December 31, 1995; chart data served to validate automated data. The complications with the highest sensitivity determined by a diagnosis in the medical records identified within +/-60 days of the database date were myocardial infarction (95.2%); amputation (94.4%); ischemic heart disease (90.3%); stroke (91.2%); osteomyelitis (79.2%); and retinal detachment, vitreous hemorrhage, and vitrectomy (73.5%). With the exception of amputation (82.9%), positive predictive value was low when based on a diagnosis identified within +/-60 days of the database date but increased with relaxation of the time constraints to include confirmation of the condition at any time during 1993-1995: ulcers (88.5%); amputation (85.4%); and retinal detachment, vitreous hemorrhage and vitrectomy (79.8%). Automated data are useful for ascertaining potential cases of some diabetic complications but require confirmatory evidence when they are to be used for research purposes.

Episodes of care: theoretical frameworks versus current operational realities

BACKGROUND

Fundamental changes in the structure of the health care industry have stimulated the need for improved definitions of output and for better methods of organizing utilization data into appropriate units. Although the "episode of care" concept has existed since the 1960s, its recognition as integral to the management of health care cost and utilization is relatively recent. Conceptually, episodes of care represent a meaningful unit of analysis for assessing the full range of primary and specialty services provided in treating a particular health problem. Proprietary episode software grouper products are currently being used by health care organizations for the purposes of provider profiling, clinical benchmarking, disease management, and quality measurement.

DESCRIPTION OF EPISODE GROUPER SOFTWARE PRODUCTS

Four episode grouper products are described that use a computerized approach for developing episodes of care from administrative data. They are compared on several characteristics, including purpose, case-mix adjustment, comprehensiveness, and clinical flexibility. Their differences in episode construction, such as how the start points and endpoints of an episode are defined, are also delineated.

CONCLUSIONS

Episode groupers are critical to the analysis of health care delivery, since they focus on the entire process of care. Although all the groupers reviewed have many strengths, much developmental work still needs to occur in order to standardize the measurement and operationalization of episodes of care as units of analysis. Furthermore, until the data sources used are more valid and reliable, they will at best remain gross screening measures of quality.

Developing a tool for analyzing medical care utilization of adult asthma patients in indemnity and managed care plans: can an episodes of care framework be used?

We evaluated the health care resource utilization between adult asthma patients in managed care organizations with those in indemnity plans using an episodes of care methodology. We also examined the importance of risk adjustment in explaining variation in resource utilization. Episodes were constructed using private insurance claims from 1992 to 1993. Bivariate and multivariate analyses were used to examine differences between managed care and indemnity plans on episode severity, resource utilization, and outcome measures (asthma-related hospitalizations and emergency room visits). Managed care plans showed higher resource utilization in terms of services and payments per episode compared with fee-for-service plans. Financial incentives to both providers and patients may have contributed to the higher utilization among managed care patients. An episodes of care methodology has potential to serve as a cost-effective "tool" in analyzing trends in medical care utilization within a health care plan. Validation of this methodology is necessary, however, before it can be used to compare trends in utilization across health plans.