Comparison Of Administrative Data Research Articles

Estimating patterns in the under-reporting of suicide deaths in India: comparison of administrative data and Global Burden of Disease Study estimates, 2005–2015

BackgroundIt has been proposed that the National Crime Records Bureau (NCRB), which reports suicides in India, differentially underenumerates suicides by geographic and demographic factors. We assessed the extent of potential...

A comparison of administrative data and quality improvement registries for abdominal aortic aneurysm repair

ObjectiveDatabases are essential in evaluating surgical outcomes and gauging the implementation of new techniques. However, there are important differences in how data from administrative databases and surgical quality improvement (QI) registries are collected and interpreted. Therefore, we aimed to compare trends, demographics, and outcomes of open and endovascular abdominal aortic aneurysm (AAA) repair in an administrative database and two QI registries. MethodsWe identified patients undergoing open and endovascular repair of intact and ruptured AAAs between 2012 and 2015 within the National Inpatient Sample (NIS), the National Surgical Quality Improvement Program (NSQIP), and the Vascular Quality Initiative (VQI). We described the differences and trends in overall AAA repairs for each data set. Moreover, patient demographics, comorbidities, mortality, and complications were compared between the data sets using Pearson χ2 test. ResultsA total of 140,240 NIS patients, 10,898 NSQIP patients, and 26,794 VQI patients were included. Ruptured repairs composed 8.7% of NIS, 11% of NSQIP, and 7.9% of VQI. Endovascular aneurysm repair (EVAR) rates for intact repair (range, 83%-84%) and ruptured repair (range, 51%-59%) were similar in the three databases. In general, rates of comorbidities were lower in NIS than in the QI registries. After intact EVAR, in-hospital mortality rates were similar in all three databases (NIS 0.8%, NSQIP 1.0%, and VQI 0.8%; P = .06). However, after intact open repair and ruptured repair, in-hospital mortality was highest in NIS and lowest in VQI (intact open: NIS 5.4%, NSQIP 4.7%, and VQI 3.5% [P < .001]; ruptured EVAR: NIS 24%, NSQIP 20%, and VQI 16% [P < .001]; ruptured open: NIS 36%, NSQIP 31%, and VQI 26% [P < .001]). After stratification by intact and ruptured presentation and repair strategy, several discrepancies in morbidity rates remained between the databases. Overall, the number of cases in NSQIP represents 7% to 8% of the repairs in NIS, and the number of cases in VQI grew from 12% in 2012 to represent 23% of the national sample in 2015. ConclusionsNIS had the largest number of patients as it represents the nationwide experience and is an essential tool to evaluate trends over time. The lower in-hospital mortality seen in NSQIP and VQI questions the generalizability of the studies that use these QI registries. However, with a growing number of hospitals engaging in granular QI initiatives, these QI registries provide a valuable resource to potentially improve the quality of care provided to all patients.

Not All Databases Are Created Equal, a Comparison of Administrative Data and Quality Improvement Registries for Abdominal Aortic Aneurysm Repair

Databases are essential in evaluating surgical outcomes. However, there are important differences in how data from administrative databases and surgical quality improvement (QI) registries are collected and interpreted. Therefore, we aim to compare trends, demographics, and outcomes of open and endovascular aortic abdominal aneurysm (AAA) repair in an administrative database and two QI registries.

Comparison of administrative data and the American College of Surgeons National Surgical Quality Improvement Program data in a New South Wales Hospital.

The National Surgical Quality Improvement Program (NSQIP) is widely used in North America for benchmarking. In 2015, NSQIP was introduced to four New South Wales public hospitals. The aim of this study is to investigate the agreement between NSQIP and administrative data in the Australian setting; to compare the performance of models derived from each data set to predict 30-day outcomes. The NSQIP and administrative data variables were mapped to select variables available in both data sets where coding may be influenced by interpretation of the clinical information. These were compared for agreement. Logistic regression models were fitted to estimate the probability of adverse outcomes within 30 days. Models derived from NSQIP and administrative data were compared by receiver operating characteristic curve analysis. A total of 2240 procedures over 21 months had matching records. Functional status demonstrated poor agreement (kappa 0.02): administrative data recorded only one (1%) patient with partial- or total-dependence as recorded by NSQIP data. The American Society of Anesthesiologists class demonstrated excellent agreement (kappa 0.91). Other perioperative variables demonstrated poor to fair agreement (kappa 0.12-0.61). Predictive model based on NSQIP data was excellent at predicting mortality but was less accurate for complications and readmissions. The NSQIP model was better in predicting mortality and complications (receiver operating characteristic curve 0.93 versus 0.87; P = 0.029 and 0.71 versus 0.64; P = 0.027). There is poor agreement between NSQIP data and administrative data. Predictive models associated with NSQIP data were more accurate at predicting surgical outcomes than those from administrative data. To drive quality improvement in surgery, high-quality clinical data are required and we believe that NSQIP fulfils this function.

36th International Symposium on Intensive Care and Emergency Medicine : Brussels, Belgium. 15-18 March 2016.

36th International Symposium on Intensive Care and Emergency Medicine : Brussels, Belgium. 15-18 March 2016.

Measuring compliance to capecitabine in colon cancer: A comparison of administrative data and medical records.

531 Background: Studies of administrative data indicate that compliance with oral anti-neoplastic agents, including capecitabine, is suboptimal. Using a population-based cohort of colon cancer, our aim was to evaluate the proportion of patients deemed non-compliant that actually had a reasonable indication for not adhering to their prescribed oral systemic therapy. Methods: Consecutive patients diagnosed with stage III colon cancer from 2008 to 2010, referred to any 1 of 5 regional cancer centers in British Columbia, Canada, and who initiated at least 1 cycle of oral adjuvant capecitabine within 12 weeks of curative resection were reviewed. Administrative data from the provincial oncology pharmacy database were analyzed to assess for non-compliance, which was defined as any prescription refill delays of &gt;/=1 week from the end date of the preceding cycle. Electronic medical records were abstracted to examine the factors, if any, which contributed to non-compliance. We compared administrative vs. medical record data to determine the level of concordance and the appropriateness of non-compliance. Results: We included 752 patients: median age was 70 years (IQR 35-87), 56% were men, and 86% had ECOG 0/1. Administrative data showed that 413 patients were non-compliant: 230 (56%) and 183 (44%) had 1 and &gt;/=2 late refills, respectively. In this group, a total of 2095 cycles of capecitabine were delivered among which 688 (33%) treatment delays occurred. No differences in baseline characteristics were observed between individuals who were compliant and those who were not (all p&gt;0.05). Of the 688 prescription delays that were ascertained from administrative data, medical records demonstrated that 30 (4%) were misclassified. From the remainder, the majority were attributable to valid reasons, including: 339 (49%) toxicities necessitating time off therapy; 170 (25%) physician discretion; 40 (6%) travel requiring adjustment to treatment schedule; and 6 (1%) patient refusal. Only 103 (15%) cases were considered as true non-compliance. Conclusions: Using administrative data alone to measure oral drug compliance without corroborating with medical records may overestimate the degree of non-adherence.

Invited Commentary

In the current health care environment, there is a seemingly endless amount of data with which one can assess performance, identify patterns, compare processes, and hopefully improve outcomes. We have powerful software and tools to help us use data in ways not imagined when the data were first collected. Yet, to use data wisely, one must remember the words of 19th century mathematician Charles Babbage [1Babbage C. Passages from the life of a philosopher. Longman and Company, London1864Google Scholar]:On two occasions I have been asked, “Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?” I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question. In other words, as the old adage of computing goes garbage in, garbage out. In this article, Pasquali and colleagues [2Pasquali S.K. Peterson E.D. Jacobs J.P. et al.Differential case ascertainment in clinical registry versus administrative data and impact on outcomes assessment in pediatric cardiac operations.Ann Thorac Surg. 2013; 95: 197-203Abstract Full Text Full Text PDF PubMed Scopus (84) Google Scholar] assess the quality of the figures we put into our machines, using an example in pediatric heart surgery. In a comparison of administrative data from the Pediatric Health Information Systems (PHIS) database with clinical data from the Society for Thoracic Surgeons Congenital Heart Surgery (STS-CHS) database, agreement between the two databases was poor. Furthermore, the administrative database had a rather low positive predictive value for identifying true cases of pediatric heart surgery. As the authors indicate, the finding that administrative data fare poorly as compared to clinical data with regards to accuracy of diagnosis and procedure is not novel. However, what is noteworthy about these findings is the effect that these inaccuracies may have on assessing and comparing outcomes. The administrative data substantially overestimated the mortality for some operations and underestimated the mortality for others. For a system in which hospitals are often compared based on outcomes determined by administrative data, these results should be eye-opening. Where should we go from here? Rather than focus on the inadequacies of administrative data demonstrated in this article, I believe that we should recognize the paradigm shift that these findings present. No database is perfect; each has its own set of strengths and weaknesses. By merging databases, however, we can maximize the strengths while minimizing the weaknesses. In the absence of global unique identifiers, this is not an easy task. Yet the opportunity (and challenge) before us lies in the ability to successfully link databases such as PHIS and STS-CHS with a myriad of other databases in critical care, anesthesia, catheterization, electrophysiology, federal agencies, and more. Only by taking advantage of the strengths that the various databases offer will we be able to put the right figures into the machine so that the right answers may come out. Differential Case Ascertainment in Clinical Registry Versus Administrative Data and Impact on Outcomes Assessment for Pediatric Cardiac OperationsThe Annals of Thoracic SurgeryVol. 95Issue 1PreviewAdministrative datasets are often used to assess outcomes and quality of pediatric cardiac programs; however their accuracy regarding case ascertainment is unclear. We linked patient data (2004–2010) from the Society of Thoracic Surgeons Congenital Heart Surgery (STS-CHS) Database (clinical registry) and the Pediatric Health Information Systems (PHIS) database (administrative database) from hospitals participating in both to evaluate differential coding/classification of operations between datasets and subsequent impact on outcomes assessment. Full-Text PDF

Advantages and limitations of using national administrative data on obstetric blood transfusions to estimate the frequency of obstetric hemorrhages

Obstetric hemorrhages are a frequent cause of maternal death all over the world, but are not routinely monitored. Health systems administrative databases could be used for this purpose, but data quality needs to be assessed. Using blood transfusion data recorded in administrative databases to estimate the frequency of obstetric hemorrhages. Research design A population-based study. Subjects Validation sub-sample: all mothers who gave birth in a French region in 2006-07 (35 123 pregnancies). Main study: all mothers who gave birth in France in 2006-07 (1 629 537 pregnancies). Linkage and comparison of administrative data on blood transfusions with data from the French blood agency ('gold standard'), and, based on this validation, the construction of a multivariable regression model to correct the number of pregnant women identified as having received a transfusion in the national administrative database. The blood transfusion rate observed in the gold standard was 7.12‰. The sensitivity of the administrative data was estimated at 66.3% and the positive predictive value at 91.3%. The estimated total number of pregnant women who received blood transfusions in France in 2006-07 was 10 941 (6.71‰). The administrative data, available in most countries, can be used to estimate the frequency of obstetric hemorrhages.

Comparison of administrative data with the Association of Coloproctology of Great Britain and Ireland (ACPGBI) colorectal cancer database

This study compared case volume and operative mortality from surgery for colorectal cancer in England derived from Hospital Episode Statistics (HES) with the Association of Coloproctology of Great Britain and Ireland (ACPGBI) colorectal cancer database. Data extracted from HES records for 2001-2002 for patients undergoing one of seven procedures for colorectal cancer were compared with those from the ACPGBI database. The primary endpoint was a 30-day post-operative mortality. 16,346 patients from HES were compared with 7,635 from the ACPGBI database. For trusts with patients in both databases, HES reported 12% more procedures than ACPGBI (7,516 vs 6,617). Records of anterior resection revealed reasonable agreement between HES and the ACPGBI databases (difference, 2%). By trust, the overall correlation between the reported procedures was 0.660. Reported crude mortality was inconsistent between the databases, with mortality from abdominoperineal excision of rectum showing the poorest correlation (r = 0.253). Overall, agreement between reported caseload and mortality was reasonable at a national, but not hospital, level. Investigation of differences between the two databases at unit level may help to detect under reporting of cases. The combination of data from both sources could be used to develop an enhanced system for monitoring outcomes from colorectal surgery in England.

Comparison of Administrative Data and Medical Records to Measure the Quality of Medical Care Provided to Vulnerable Older Patients

Administrative data are used to determine performance for publicly reported in health plan "report cards," accreditation status, and reimbursement. However, it is unclear how performance based on administrative data and medical records compare. We compared applicability, eligibility, and performance on 182 measures of health care quality using medical records and administrative data during a 13-month period for a random sample of 399 vulnerable older patients enrolled in managed care. Of 182 quality indicators (QIs) spanning 22 conditions, 145 (80%) were applicable only to medical records and 37 (20%) to either medical records or administrative data. Among 48 QIs specific to geriatric conditions, all were applicable to medical records; 2 of these also were applicable to administrative data. Eligibility for the 37 QIs that were applicable to both medical records and administrative data was similar for both data sources (94% agreement, kappa = 0.74). With the use of medical records, 152 of the 182 the QIs that were applicable to medical records were triggered and yielded an overall performance of 55%. Using administrative data, 30 of the 37 QIs that were applicable to administrative data were triggered and yielded overall performance of 83% (P < 0.05 vs. medical records). Restricting to QIs applicable to both data sources, overall performance was 84% and 83% (P = 0.21) for medical records and administrative data, respectively. The number and spectrum of QIs that can be measured for vulnerable elderly patients is far greater for medical records than for administrative data. Although summary estimates of health care quality derived from administrative data and medical records do not differ when using identical measures, summary scores from these data sources vary substantially when the totality of care that can be measured by each data source is measured.