Assessing the validity of diagnostic information in administrative health care utilization data: experience in Saskatchewan

Epidemiological and Clinical Features in Very Old Men and Women (≥80 Years) Hospitalized with Aortic Stenosis in Spain, 2016-2019: Results from the Spanish Hospital Discharge Database

(1) Background: The aging population poses challenges for hospital systems. Aortic stenosis is among the most frequent diseases in very old patients. The aim of this study was to describe gender and age differences in the clinical characteristics of very old patients hospitalized with aortic stenosis (AoS) in Spain from 2016 to 2019. (2): Methods: A retrospective observational study analyzing data from the national surveillance system for hospital data. Variables analyzed were age group, sex, length of stay, deaths, and comorbidity. (3) Results: The analysis included 46,967 discharges. Altogether, 7.6% of the admissions ended in death. The main reason for admission was heart failure (34.3%), and this increased with age (80−84 years: 26% versus 95−99 years: 56.6%; p < 0.001). The main treatment procedure was the transcatheter aortic valve replacement (12.7%), performed in 14.3% of patients aged 80−84 versus 0.5% in patients aged 95−99 (p < 0.001). In the multivariable analysis, women were admitted with more comorbidities (odds ratio [OR] 1.22, 95% confidence interval [CI] 1.06−1.20). Mortality was similar, albeit women were admitted less for syncope (OR 0.83, 95% CI 0.74−0.93). Women also underwent fewer coronary catheterizations (OR 0.81, 95% CI 0.77−0.87) and echocardiograms (OR 0.96, 95% CI 0.94−0.98). (4) Conclusions: Aortic stenosis leads to a high number of hospital admissions. Women with AoS presented more heart failure and less cardiovascular pathology than men. Also, women are admitted with fewer episodes of syncope and have fewer ultrasounds and catheterizations.

The impact of the demographic shift on limb amputation incidence in Saskatchewan, Canada, 2006–2019

Background

Changing demographics in a population may have an inevitable influence on disease incidence including limb amputation. However, the extent to which these changes affect limb amputation (LA) is unknown. Understanding the impact of changing demographics on LA would provide the best opportunity to plan for the future. We assessed the impact of changes in age and sex on limb amputation in Saskatchewan between 2006 and 2019.

Methods

Retrospective linked Saskatchewan’s LA cases, and demographic characteristics and residents population from 2006–2019 was used. The amputation rate was calculated by dividing the total number of LA cases recorded each year by the annual Saskatchewan resident population and the results expressed per 100,000 populations. Furthermore, decomposition analysis was used to assess the impact of changes in age and sex on LA in a decade (2008–2017) and the Generalized Additive Model (GAM) was employed to examine the linear and non-linear effect of age.

Results

We found that in the ten years (2008–2017), the absolute LA rate difference was 9.0 per 100,000 population. Changes in age structure alone contributed 7.7% to the LA rate increase and 92.3% to changes in age-specific LA rates. The decade witnessed a marginal population difference between males and females, but the LA rate was 2.1–2.2 times higher in males than in females. The GAM revealed a non-linear relationship between LA and age, and further indicates that the risk of LA significantly increased as age increases.

Conclusions

In a decade, we found that changes in age distribution and age-specific rate substantially impacted the increase in the LA rate observed in the province. This highlights the urgent need for strategized programs to respond to these changes as both the population and diabetes, which is age-dependent and a leading cause of LA, are expected to increase in the province by 2030. As changes in population and demographic factors are inevitable, this study provides data for policy makers on the need for continuous incorporation of the shift in population in the design of future health services.

Trends of limb amputation considering type, level, sex and age in Saskatchewan, Canada 2006-2019: an in-depth assessment

BACKGROUND

Most epidemiologic reports focus on lower extremity amputation (LEA) caused specifically by diabetes mellitus. However, narrowing scope disregards the impact of other causes and types of limb amputation (LA) diminishing the true incidence and societal burden. We explored the rates of LEA and upper extremity amputation (UEA) by level of amputation, sex and age over 14 years in Saskatchewan, Canada.

METHODS

We calculated the differential impact of amputation type (LEA or UEA) and level (major or minor) of LA using retrospective linked hospital discharge data and demographic characteristics of all LA performed in Saskatchewan and resident population between 2006 and 2019. Rates were calculated from total yearly cases per yearly Saskatchewan resident population. Joinpoint regression was employed to quantify annual percentage change (APC) and average annual percent change (AAPC). Negative binomial regression was performed to determine if LA rates differed over time based on sex and age.

RESULTS

Incidence of LEA (31.86 ± 2.85 per 100,000) predominated over UEA (5.84 ± 0.49 per 100,000) over the 14-year study period. The overall LEA rate did not change over the study period (AAPC -0.5 [95% CI - 3.8 to 3.0]) but fluctuations were identified. From 2008 to 2017 LEA rates increased (APC 3.15 [95% CI 1.1 to 5.2]) countered by two statistically insignificant periods of decline (2006-2008 and 2017-2019). From 2006 to 2019 the rate of minor LEA steadily increased (AAPC 3.9 [95% CI 2.4 to 5.4]) while major LEA decreased (AAPC -0.6 [95% CI - 2.1 to 5.4]). Fluctuations in the overall LEA rate nearly corresponded with fluctuations in major LEA with one period of rising rates from 2010 to 2017 (APC 4.2 [95% CI 0.9 to 7.6]) countered by two periods of decline 2006-2010 (APC -11.14 [95% CI - 16.4 to - 5.6]) and 2017-2019 (APC -19.49 [95% CI - 33.5 to - 2.5]). Overall UEA and minor UEA rates remained stable from 2006 to 2019 with too few major UEA performed for in-depth analysis. Males were twice as likely to undergo LA than females (RR = 2.2 [95% CI 1.99-2.51]) with no change in rate over the study period. Persons aged 50-74 years and 75+ years were respectively 5.9 (RR = 5.92 [95% Cl 5.39-6.51]) and 10.6 (RR = 10.58 [95% Cl 9.26-12.08]) times more likely to undergo LA than those aged 0-49 years. LA rate increased with increasing age over the study period.

CONCLUSION

The rise in the rate of minor LEA with simultaneous decline in the rate of major LEA concomitant with the rise in age of patients experiencing LA may reflect a paradigm shift in the management of diseases that lead to LEA. Further, this shift may alter demand for orthotic versus prosthetic intervention. A more granular look into the data is warranted to determine if performing minor LA diminishes the need for major LA.

The influence of primary and subsequent limb amputation on the overall rate of limb amputation in Saskatchewan, Canada, 2006-2019: a population-based study

Background

Understanding trends in limb amputation (LA) can provide insight into the prevention and optimization of health care delivery. We examine the influence of primary (first report) and subsequent (multiple reports) limb amputation on the overall (all reports) rate of limb amputation in Saskatchewan considering amputation level.

Methods

Hospital discharged data associated with LA from 2006 to 2019 and population estimates in Saskatchewan were used. LA cases were grouped based on overall, primary, and subsequent LA and further divided by level into major (through/above the ankle/wrist) and minor (below the ankle/wrist). Incidence rates were calculated using LA cases as the numerator and resident population as the denominator. Joinpoint and negative binomial were used to analyze the trends. In addition, the top three amputation predisposing factors (APF) were described by LA groups.

Results

The rate of overall LA and primary LA remained stable (AAPC − 0.9 [95% CI − 3.9 to 2.3]) and (AAPC −1.9 [95% CI −4.2 to 0.4]) respectively, while the rate of subsequent LA increased 3.2% (AAPC 3.2 [95% CI 3.1 to 9.9]) over the 14-year study period. The rate of overall major LA declined 4.6% (AAPC − 4.6 [95% CI −7.3 to −1.7]) and was largely driven by the 5.9% decline in the rate of primary major LA (AAPC − 5.9 [95% CI − 11.3 to –0.2]). Subsequent major LA remained stable over the study period (AAPC −0.4 [95% CI − 6.8 to 6.5]). In contrast, the overall rate of minor LA increased 2.0% (AAPC 2.0 [95% CI 1.0 to 2.9]) over the study period which was largely driven by a 9.6% increase in the rate of subsequent minor LA (AAPC 9.6 [95% CI 4.9 to 14.4]). Primary minor LA rates remained stable over the study period (AAPC 0.6 [95% CI − 0.2 to 1.5]). The study cohorts were 1.3-fold greater risk of minor LA than major LA. Diabetes mellitus (DM) was the leading APF representing 72.8% of the cohort followed by peripheral vascular disease (PVD) and trauma with 17.1 and 10.1% respectively. Most (86.7%) of subsequent LA were performed on people with DM.

Conclusions

Overall LA rates remained stable over the study period with declining rates of major LA countered by rising rates of minor LA. Minor LA exceeded major LA with the largest rate increase identified in subsequent minor LA. Diabetes was the greatest APF for all LA groups. This rising rate of more frequent and repeated minor LA may reflect changing intervention strategies implemented to maintain limb function. The importance of long-term surveillance to understand rates of major and minor LA considering primary and subsequent intervention is an important step to evaluate and initiate prevention and limb loss management programs.

Incidence and trends of limb amputation in first nations and general population in Saskatchewan, 2006-2019

BACKGROUND

There is conflicting evidence whether limb amputation (LA) disproportionately affects indigenous populations. To better understand this disparity, we compared the LA incidence rate between First Nations persons registered under the Indian Act of Canada (RI) and the general population (GP) in Saskatchewan.

METHODS

We used Saskatchewan's retrospective administrative data containing hospital discharge LA cases, demographic characteristics (age and sex), and residents population reported in the database stratified by RI and GP from 2006-2019. The LA cases for each stratified group were first disaggregated into three broad categories: overall LA (all reported LA), primary LA (first reported LA), and subsequent LA (revision or contralateral LA), with each category further split into the level of amputation defined as major amputation (through/above the ankle/wrist joint) and minor amputation (below the ankle/wrist joint). LA rates were calculated using LA cases as the numerator and resident population as the denominator. Joinpoint and negative binomial regressions were performed to explore the trends further.

RESULTS

Overall, there were 1347 RI and 4520 GP LA cases reported in Saskatchewan from 2006-2019. Primary LA made up approximately 64.5% (869) of RI and 74.5% (3369) of GP cases, while subsequent LA constituted 35.5% (478) of RI and 25.5% (1151) of GP cases. The average age-adjusted LA rate was 153.9 ± 17.3 per 100,000 in the RI cohort and 31.1 ± 2.3 per 100,000 in the GP cohort. Overall and primary LA rates for the GP Group declined 0.7% and 1.0%, while subsequent LA increased 0.1%. An increased LA rate for all categories (overall 4.9%, primary 5.1%, and subsequent 4.6%) was identified in the RI group. Overall, minor and major LA increased by 6.2% and 3.3%, respectively, in the RI group compared to a 0.8% rise in minor LA and a 6.3% decline in major LA in the GP group. RI females and males were 1.98-1.66 times higher risk of LA than their GP counterparts likewise, RI aged 0-49 years and 50+ years were 2.04-5.33 times higher risk of LA than their GP cohort. Diabetes mellitus (DM) was the most prevalent amputation predisposing factor in both groups with 81.5% of RI and 54.1% of GP diagnosed with DM. Also, the highest proportion of LA was found in the lowest income quintile for both groups (68.7% for RI and 45.3% for GP).

CONCLUSION

Saskatchewan's indigenous individuals, specifically First Nations persons registered under the Indian Act of Canada, experience LA at a higher rate than the general population. This disparity exists for all variables examined, including overall, primary, and subsequent LA rates, level of amputation, sex, and age. Amplification of the disparities will continue if the rates of change maintain their current trajectories. These results underscore the need for a better understanding of underlying causes to develop a targeted intervention in these groups.

Time trends in hospital discharges in patients aged 85 years and older in Spain: data from the Spanish National Discharge Database (2000-2015)

Background

The aging population is an increasing concern in Western hospital systems. The aim of this study was to describe the main characteristics and hospitalization patterns in inpatients aged 85 years or more in Spain from 2000 to 2015.

Methods

Retrospective observational study analyzing data from the minimum basic data set, an administrative registry recording each hospital discharge in Spain since 1997. We collected administrative, economic and clinical data for all discharges between 2000 and 2015 in patients aged 85 years and older, reporting results in three age groups and four time periods to assess differences and compare trends.

Results

There were 4,387,326 discharges in very elderly patients in Spain from 2000 to 2015, representing 5.32% of total discharges in 2000–2003 and 10.42% in 2012–2015. The pace of growth was faster in older age groups, with an annual percentage increase of 6% in patients aged 85–89 years, 7.79% in those aged 90–94 years, and 8.06% in those aged 95 and older. The proportion of men also rose (37.30 to 39.70%, p < 0.001).

The proportion of patients that died during hospital admission decreased from 14.64% in 2000–2003 to 13.83% in 2012–2015 (p < 0.001), and mean length of stay from 9.98 days in 2000–2003 to 8.34 days in 2012–2015.

Some of the most frequent primary diagnoses became even more frequent relative to the total number of primary diagnoses, such as heart failure (7.84 to 10.62%), pneumonia (6.36 to 7.36%), other respiratory diseases (3.87 to 8.49%) or other alterations of urinary tract (3.08 to 5.20%). However, there was a relative decrease in the proportion of femoral neck fractures (8.07 to 6.77%), neoplasms (7.65 to 7.34%), ischemic encephalopathy (6.97 to 5.85%), COPD (4.23 to 3.15%), ischemic cardiomyopathy (4.20 to 8.49%) and cholelithiasis (3.07 to 3.28%).

Conclusions

Discharges in the very elderly population are increasing in both relative and absolute terms in Spanish hospitals. Within this group, discharged patients are getting older and more frequently male. The mean length of stay and the proportion of patients that died during hospital admission are decreasing. Acute-on-chronic organ diseases, neoplasms, acute cardiovascular diseases, and infections are the most common causes of discharge.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-021-02335-2.

Validity of Algorithms for Identification of Individuals Suffering from Chronic Noncancer Pain in Administrative Databases: A Systematic Review

BACKGROUND

Secondary analysis of health administrative databases is indispensable to enriching our understanding of health trajectories, health care utilization, and real-world risks and benefits of drugs among large populations.

OBJECTIVES

This systematic review aimed at assessing evidence about the validity of algorithms for the identification of individuals suffering from nonarthritic chronic noncancer pain (CNCP) in administrative databases.

METHODS

Studies reporting measures of diagnostic accuracy of such algorithms and published in English or French were searched in the Medline, Embase, CINAHL, AgeLine, PsycINFO, and Abstracts in Social Gerontology electronic databases without any dates of coverage restrictions up to March 1, 2018. Reference lists of included studies were also screened for additional publications.

RESULTS

Only six studies focused on commonly studied CNCP conditions and were included in the review. Some algorithms showed a ≥60% combination of sensitivity and specificity values (back pain disorders in general, fibromyalgia, low back pain, migraine, neck/back problems studied together). Only algorithms designed to identify fibromyalgia cases reached a ≥80% combination (without replication of findings in other studies/databases).

CONCLUSIONS

In summary, the present investigation informs us about the limited amount of literature available to guide and support the use of administrative databases as valid sources of data for research on CNCP. Considering the added value of such data sources, the important research gaps identified in this innovative review provide important directions for future research. The review protocol was registered with PROSPERO (CRD42018086402).

Oral anticoagulation use by patients with atrial fibrillation in Germany

Atrial fibrillation (AF) is the most common significant cardiac rhythm disorder. Oral anticoagulation (OAC) is recommended by guidelines in the presence of a moderate to high risk of stroke. Based on an analysis of claims-based data, the aim of this contribution is to quantify the stroke-risk dependent OAC utilisation profile of German AF patients as well as the possible causes and the associated clinical outcomes of OAC under-use. Our data set was derived from two large mandatory German medical insurance funds. Risk stratification of patients was based on the CHADS2-score and the CHA2DS2-VASc-score. Two different scenarios were constructed to deal with factors potentially disfavouring OAC use. Causes of OAC under-use and its clinical consequences were analysed using multivariate analysis. Observation year was 2008. A total of 183,448 AF patients met the inclusion criteria. This represents an AF prevalence of 2.21%. The average CHADS2-score was 2.8 (CHA2DS2-VASc-score: 4.3). On between 40.5 and 48.7% of the observed patient-days, there was no antithrombotic protection by OAC, other anticoagulants or aspirin. Older female patients with a high number of comorbidities had a higher risk of OAC under-use. Patients who had already experienced a thromboembolic event had a lower risk of OAC under-use. In the observation year, 3,367 patients experienced a stroke (incidence rate 1.8%). In our multi-level Poisson random effects estimate, OAC use decreases the stroke rate by almost 80% (IRR 0.236). In conclusion, OAC under-use is widespread in the German market. It is associated with severe clinical consequences.

Drug survival of second biological DMARD therapy in patients with rheumatoid arthritis: a retrospective non-interventional cohort analysis

Background

Since persistence to first biological disease modifying anti-rheumatic drugs (bDMARDs) is far from ideal in rheumatoid arthritis (RA) patients, many do receive a second and/or third bDMARD treatment. However, little is known about treatment persistence of the second-line bDMARD and it is specifically unknown whether the mode of action of such a treatment is associated with different persistence rates. We aimed to assess discontinuation-, re-initiation- or continuation-rates of a 2nd bDMARD therapy as well as switching-rates to a third biological DMARD (3rd bDMARD) therapy in RA patients.

Method

Analysis was based on German claims data (2010–2013). Patients were included if they had received at least one prescription for an anti-TNF and at least one follow-up prescription of a 2nd bDMARD different from the first anti-TNF. Patient follow-up started on the date of the first prescription for the 2nd bDMARD and lasted for 12 months or until a patient’s death.

Results

2667 RA patients received at least one anti-TNF prescription. Of these, 451 patients received a second bDMARD (340 anti-TNF, mean age 52.6 years; 111 non-anti-TNF, mean age 55.9 years).

During the follow-up, 28.8% vs. 11.7% of the 2nd anti-TNF vs. non-anti-TNF patients (p < 0.001) switched to a 3rd bDMARD; 14.1% vs. 19.8% (p = 0.179) discontinued without re-start; 3.8% vs.1.8% (p = 0.387) re-started and 53.5 vs. 66.7% (p < 0.050) continued therapy. Patients in the non-anti-TNF group demonstrated longer drug survival (295 days) than patients in the anti-TNF group (264 days; p = 0.016).

Independent variables associated with earlier discontinuation (including re-start) or switch were prescription of an anti-TNF as 2nd bDMARD (HR = 1.512) and a higher comorbidity level (CCI, HR = 1.112), whereas previous painkiller medication (HR = 0.629) was associated with later discontinuation or switch.

Conclusions

Only 56.8% of RA patients continued 2nd bDMARD treatment after 12 months; 60% if re-start was included. Non-anti-TNF patients had a higher probability of continuing 2nd bDMARD therapy.

Electronic supplementary material

The online version of this article (doi:10.1186/s12891-017-1684-0) contains supplementary material, which is available to authorized users.

Real life anticoagulation treatment of patients with atrial fibrillation in Germany: extent and causes of anticoagulant under-use

Oral anticoagulation (OAC) with either new oral anticoagulants (NOACs) or Vitamin-K antagonists (VKAs) is recommended by guidelines for patients with atrial fibrillation (AF) and a moderate to high risk of stroke. Based on a claims-based data set the aim of this study was to quantify the stroke-risk dependent OAC utilization profile of German AF patients and possible causes of OAC under-use. Our claims-based data set was derived from two German statutory health insurance funds for the years 2007-2010. All prevalent AF-patients in the period 2007-2009 were included. The OAC-need in 2010 was assumed whenever a CHADS2- or CHA2DS2-VASC-score was >1 and no factor that disfavored OAC use existed. Causes of OAC under-use were analyzed using multivariate logistic regression. 108,632 AF-prevalent patients met the inclusion criteria. Average age was 75.43 years, average CHA2DS2-VASc-score was 4.38. OAC should have been recommended for 56.1/62.9 % of the patients (regarding factors disfavouring VKA/NOAC use). For 38.88/39.20 % of the patient-days in 2010 we could not observe any coverage by anticoagulants. Dementia of patients (OR 2.656) and general prescription patterns of the treating physician (OR 1.633) were the most important factors increasing the risk of OAC under-use. Patients who had consulted a cardiologist had a lower risk of being under-treated with OAC (OR 0.459). OAC under-use still seems to be one of the major challenges in the real-life treatment of AF patients. Our study confirms that both patient/disease characteristics and treatment environment/general prescribing behaviour of physicians may explain the OAC under-use in AF patients.

Agreement between administrative data and the Resident Assessment Instrument Minimum Dataset (RAI-MDS) for medication use in long-term care facilities: a population-based study

Background

Prescription medication use, which is common among long-term care facility (LTCF) residents, is routinely used to describe quality of care and predict health outcomes. Data sources that capture medication information, which include surveys, medical charts, administrative health databases, and clinical assessment records, may not collect concordant information, which can result in comparable prevalence and effect size estimates. The purpose of this research was to estimate agreement between two population-based electronic data sources for measuring use of several medication classes among LTCF residents: outpatient prescription drug administrative data and the Resident Assessment Instrument Minimum Data Set (RAI-MDS) Version 2.0.

Methods

Prescription drug and RAI-MDS data from the province of Saskatchewan, Canada (population 1.1 million) were linked for 2010/11 in this cross-sectional study. Agreement for anti-psychotic, anti-depressant, and anti-anxiety/hypnotic medication classes was examined using prevalence estimates, Cohen’s κ, and positive and negative agreement. Mixed-effects logistic regression models tested resident and facility characteristics associated with disagreement.

Results

The cohort was comprised of 8,866 LTCF residents. In the RAI-MDS data, prevalence of anti-psychotics was 35.7%, while for anti-depressants it was 37.9% and for hypnotics it was 27.1%. Prevalence was similar in prescription drug data for anti-psychotics and anti-depressants, but lower for hypnotics (18.0%). Cohen’s κ ranged from 0.39 to 0.85 and was highest for the first two medication classes. Diagnosis of a mood disorder and facility affiliation was associated with disagreement for hypnotics.

Conclusions

Agreement between prescription drug administrative data and RAI-MDS assessment data was influenced by the type of medication class, as well as selected patient and facility characteristics. Researchers should carefully consider the purpose of their study, whether it is to capture medication that are dispensed or medications that are currently used by residents, when selecting a data source for research on LTCF populations.

Electronic supplementary material

The online version of this article (doi:10.1186/s12877-015-0023-2) contains supplementary material, which is available to authorized users.

Population-based patterns of prescription androgen use, 1976-2008

PURPOSE

Prescription testosterone (T) has limited approved medical indications and is a controlled substance in Canada. Utilization studies in other Westernized countries have revealed sharp increases in T use in recent years. We examined medical use of androgens, including T, over a ≥30-year period among adult (18+) men in a population-based study set in a Canadian juridisdiction of universal health care.

METHODS

Analyses were based on data from electronic records of dispensed prescriptions during 1976-2008 in Saskatchewan, Canada. All formulations of androgens listed in the provincial formulary (oral and injectable) were included. We examined demographics of users, androgen types used, switching patterns, and trends in the annual rate of use over time.

RESULTS

There were 11 521 androgen users who were followed for an average of 11.8 years. Overall, 11 types of androgens were used, and there were 86 812 dispensing events. The mean age at first use was 56.4 years (median: 58). Men had 7.5 prescription dispensing events on average (median: 2). The most commonly used formulations were methyl-T (36.2% of users) followed by T-enanthate (32.5%), T-cypionate (22.3%), and T-undecanoate (20.0%). Most users (82%) did not switch among androgen types. The annual rate of use varied substantially over time, with a marked increase observed from 1994 to 1999 and a decrease from 2000 to 2008.

CONCLUSIONS

Androgen users were largely middle aged and had relatively few dispensings. We hypothesize that observed secular trends in androgen use may align with drug treatment pattern changes for erectile dysfunction, including the advent of phosphodiesterase type 5 inhibitors.

Validity of the RAI-MDS for ascertaining diabetes and comorbid conditions in long-term care facility residents

BACKGROUND

This study assessed the validity of the Resident Assessment Instrument Minimum Data Set (RAI-MDS) Version 2.0 for diagnoses of diabetes and comorbid conditions in residents of long-term care facilities (LTCFs).

METHODS

Hospital inpatient, outpatient physician billing, RAI-MDS, and population registry data for 1997 to 2011 from Saskatchewan, Canada were used to ascertain cases of diabetes and 12 comorbid conditions. Prevalence estimates were calculated for both RAI-MDS and administrative health data. Sensitivity, specificity, and positive and negative predictive values (PPV and NPV) were calculated using population-based administrative health data as the validation data source. Cohen's κ was used to estimate agreement between the two data sources.

RESULTS

23,217 LTCF residents were in the diabetes case ascertainment cohort. Diabetes prevalence was 25.3% in administrative health data and 21.9% in RAI-MDS data. Overall sensitivity of a RAI-MDS diabetes diagnoses was 0.79 (95% CI: 0.79, 0.80) and the PPV was 0.92 (95% CI: 0.91, 0.92), when compared to administrative health data. Sensitivity of the RAI-MDS for ascertaining comorbid conditions ranged from 0.21 for osteoporosis to 0.92 for multiple sclerosis; specificity was high for most conditions.

CONCLUSIONS

RAI-MDS clinical assessment data are sensitive to ascertain diabetes cases in LTCF populations when compared to administrative health data. For many comorbid conditions, RAI-MDS data have low validity when compared to administrative data. Risk-adjustment measures based on these comorbidities might not produce consistent results for RAI-MDS and administrative health data, which could affect the conclusions of studies about health outcomes and quality of care across facilities.

Deriving comorbidities from medical records using natural language processing

Extracting comorbidity information is crucial for phenotypic studies because of the confounding effect of comorbidities. We developed an automated method that accurately determines comorbidities from electronic medical records. Using a modified version of the Charlson comorbidity index (CCI), two physicians created a reference standard of comorbidities by manual review of 100 admission notes. We processed the notes using the MedLEE natural language processing system, and wrote queries to extract comorbidities automatically from its structured output. Interrater agreement for the reference set was very high (97.7%). Our method yielded an F1 score of 0.761 and the summed CCI score was not different from the reference standard (p=0.329, power 80.4%). In comparison, obtaining comorbidities from claims data yielded an F1 score of 0.741, due to lower sensitivity (66.1%). Because CCI has previously been validated as a predictor of mortality and readmission, our method could allow automated prediction of these outcomes.

Predictive performance of comorbidity measures in administrative databases for diabetes cohorts

BACKGROUND

The performance of comorbidity measures for predicting mortality in chronic disease populations and using ICD-9 diagnosis codes in administrative health data has been investigated in several studies, but less is known about predictive performance with ICD-10 data and for other health outcomes. This study investigated predictive performance of five comorbidity measures for population-based diabetes cohorts in administrative data. The objectives were to evaluate performance for: (a) disease-specific and general health outcomes, (b) data based on the ICD-9 and ICD-10 diagnoses, and (c) different age groups.

METHODS

Performance was investigated for heart attack, stroke, amputation, renal disease, hospitalization, and death in all-age and age-specific cohorts. Hospital records, physician billing claims, and prescription drug records from one Canadian province were used to identify diabetes cohorts and measure comorbidity. The data were analysed using multiple logistic regression models and summarized using measures of discrimination, accuracy, and fit.

RESULTS

In Cohort 1 (n = 29,058), for which only ICD-9 diagnoses were recorded in administrative data, the Elixhauser index showed good or excellent prediction for amputation, renal disease, and death and performed better than the Charlson index. Number of diagnoses was a good predictor of hospitalization. Similar results were obtained for Cohort 2 (n = 41,925), in which both ICD-9 and ICD-10 diagnoses were recorded in administrative data, although predictive performance was sometimes higher. For age-specific models of mortality, the Elixhauser index resulted in the largest improvement in predictive performance in all but the youngest age group.

CONCLUSIONS

Cohort age and the health outcome under investigation, but not the diagnosis coding system, may influence the predictive performance of comorbidity measure for studies about diabetes populations using administrative health data.

Performance of comorbidity measures for predicting outcomes in population-based osteoporosis cohorts

The performance of five comorbidity measures, including the Charlson and Elixhauser indices, was investigated for predicting mortality, hospitalization, and fracture outcomes in two osteoporosis cohorts defined from administrative databases. The optimal comorbidity measure depended on the outcome of interest, although overall the Elixhauser index performed well.

INTRODUCTION

Studies that use administrative data to investigate population-based health outcomes often adopt risk-adjustment models that include comorbidities, conditions that coexist with the index disease. There has been limited research about the measurement of comorbidity in osteoporotic populations. The study purpose was to compare the performance of comorbidity measures for predicting mortality, fracture, and health service utilization outcomes in two cohorts with diagnosed or treated osteoporosis.

METHODS

Administrative data were from the province of Saskatchewan, Canada. Osteoporosis cohorts were identified from diagnoses in hospital and physician data and prescriptions for osteo-protective medications using case definitions with high sensitivity or high specificity. Five diagnosis- and medication-based comorbidity measures and five 1-year outcomes, including mortality, hospitalization (two measures), osteoporotic-related fracture, and hip fracture, were defined. Performance of the comorbidity measures was assessed using the c-statistic (discrimination) and Brier score (prediction error) for multiple logistic regression models.

RESULTS

In the specific cohort (n = 9,849) for the mortality outcome, the Elixhauser index resulted in the largest improvement (8.96%) in the c-statistic and lowest Brier score compared to a model that contained demographic and socioeconomic variables, followed by the Charlson index (6.06%). For hospitalization, the number of different diagnoses resulted in the largest improvement (14.01%) in the c-statistic. The Elixhauser index resulted in significant improvements in the c-statistic for osteoporosis-related and hip fractures. Similar results were observed for the sensitive cohort (n = 28,068).

CONCLUSIONS

Recommendations about the optimal comorbidity measure will vary with the outcome under investigation. Overall, the Elixhauser index performed well.

Comparing comorbidity measures for predicting mortality and hospitalization in three population-based cohorts

Background

Multiple comorbidity measures have been developed for risk-adjustment in studies using administrative data, but it is unclear which measure is optimal for specific outcomes and if the measures are equally valid in different populations. This research examined the predictive performance of five comorbidity measures in three population-based cohorts.

Methods

Administrative data from the province of Saskatchewan, Canada, were used to create the cohorts. The general population cohort included all Saskatchewan residents 20+ years, the diabetes cohort included individuals 20+ years with a diabetes diagnosis in hospital and/or physician data, and the osteoporosis cohort included individuals 50+ years with diagnosed or treated osteoporosis. Five comorbidity measures based on health services utilization, number of different diagnoses, and prescription drugs over one year were defined. Predictive performance was assessed for death and hospitalization outcomes using measures of discrimination (c-statistic) and calibration (Brier score) for multiple logistic regression models.

Results

The comorbidity measures with optimal performance were the same in the general population (n = 662,423), diabetes (n = 41,925), and osteoporosis (n = 28,068) cohorts. For mortality, the Elixhauser index resulted in the highest c-statistic and lowest Brier score, followed by the Charlson index. For hospitalization, the number of diagnoses had the best predictive performance. Consistent results were obtained when we restricted attention to the population 65+ years in each cohort.

Conclusions

The optimal comorbidity measure depends on the health outcome and not on the disease characteristics of the study population.

Validity of The Health Improvement Network (THIN) for the study of psoriasis

BACKGROUND

Psoriasis is a common disease frequently studied in large databases. To date the validity of psoriasis information has not been established in The Health Improvement Network (THIN).

OBJECTIVES

To investigate the validity of THIN for identifying patients with psoriasis and to determine if the database can be used to determine the natural history of the disease.

METHODS

First, we conducted a cross-sectional study to determine if psoriasis prevalence in THIN is similar to expected. Second, we created a cohort of 4900 patients, aged 45-64 years, with a psoriasis diagnostic Read Code and surveyed their general practitioners (GPs) to confirm the diagnosis clinically. Third, we created models to determine if psoriasis descriptors (extent, severity, duration and dermatologist confirmation) could be accurately captured from database records.

RESULTS

Psoriasis prevalence was 1·9%, and showed the characteristic age distribution expected. GP questionnaires were received for 4634 of 4900 cohort patients (95% response rate), and psoriasis diagnoses were confirmed in 90% of patients. Duration of disease in the database showed substantial agreement with physician query (κ = 0·69). GPs confirmed that the psoriasis diagnosis was corroborated by a dermatologist in 91% of patients whose database records contained a dermatology referral code associated with a psoriasis code. We achieved good discrimination between patients with and without extensive disease based on the number of psoriasis codes received per year (area under curve = 0·8).

CONCLUSIONS

THIN is a valid data resource for studying psoriasis and can be used to identify characteristics of the disease such as duration and confirmation by a dermatologist.

Psoriasis and cardiovascular risk: strength in numbers

In this issue, Wakkee and colleagues report a self-described exploratory cohort study and conclude that psoriasis may not be an independent risk factor for ischemic heart disease (IHD) hospitalization and that there is only a slight and borderline increased risk of ischemic heart disease among psoriasis patients. This negative result should be interpreted in light of the study's limitations, the complex relationship among levels of psoriasis severity, patient age, and cardiovascular (CV) risk, and the context of the rapidly growing literature.

Cardiovascular risks of nonsteroidal antiinflammatory drugs in patients after hospitalization for serious coronary heart disease

BACKGROUND

The cardiovascular safety of individual nonsteroidal antiinflammatory drugs (NSAIDs) is highly controversial, particularly in persons with serious coronary heart disease.

METHODS AND RESULTS

We conducted a multisite retrospective cohort study of commonly used individual NSAIDs in Tennessee Medicaid, Saskatchewan Health, and United Kingdom General Practice Research databases. The cohort included 48566 patients recently hospitalized for myocardial infarction, revascularization, or unstable angina pectoris with more than 111000 person-years of follow-up. Naproxen users had the lowest adjusted rates of serious coronary heart disease (myocardial infarction, coronary heart disease death) and serious cardiovascular disease (myocardial infarction, stroke)/death from any cause, with respective incidence rate ratios (relative to NSAID nonusers) of 0.88 (95% CI, 0.66 to 1.17) and 0.91 (0.78 to 1.06). Risk did not increase with doses >or=1000 mg. Relative to NSAID nonusers, serious coronary heart disease risk increased with short term (<90 days) use for ibuprofen (1.67 [1.09 to 2.57]), diclofenac (1.86 [1.18 to 2.92]), celecoxib (1.37 [0.96 to 1.94]), and rofecoxib (1.46 [1.03 to 2.07]), but not for naproxen (0.88 [0.50 to 1.55]). Relative to naproxen, current users of diclofenac had increased risk of serious coronary heart disease (1.44 [0.96 to 2.15], P=0.076) and serious cardiovascular disease/death (1.52 [1.22 to 1.89], P=0.0002), and those of ibuprofen had increased risk of the latter end point (1.25 [1.02 to 1.53], P=0.032). Compared to naproxen in doses >or=1000 mg, serious coronary heart disease incidence rate ratios were increased for rofecoxib >25 mg (2.29 [1.24 to 4.22], P=0.008) and celecoxib >200 mg (1.61 [1.01 to 2.57], P=0.046).

CONCLUSIONS

In patients recently hospitalized for serious coronary heart disease, naproxen had better cardiovascular safety than did diclofenac, ibuprofen, and higher doses of celecoxib and rofecoxib.

The sex factor: epidemiology and management of chronic obstructive pulmonary disease in British Columbia

BACKGROUND

The prevalence and mortality of chronic obstructive pulmonary disease (COPD) in women have been predicted to overtake that of men within the next decade. These predictions are based in part on data from surveys using self-reports of a COPD diagnosis. Whether these predictions have been realized is unknown.

METHODS

The prevalence and mortality of men and women in British Columbia were compared from fiscal years 1992/1993 to 2003/2004 using administrative health services data. Case definitions for COPD were developed using International Classification of Diseases ninth and 10th revision (ICD-9/10) codes applied to medical and hospital data. Individuals 45 years and older, who had at least two physician visits or one hospitalization for specified COPD ICD-9/10 codes within a 365-day window, were considered to be cases. Cases were ascertained from 1992 to 2004.

RESULTS

In 2003/2004, men had a greater prevalence (4.7% versus 4.0% in women) and a higher all-cause mortality rate (5.4% versus 4.1% in women) than women. Both men and women with COPD had low COPD medication use (45%) and low referral for lung function testing (55%). Including the ICD-9 code for 'bronchitis, not specified as acute or chronic' (ICD-9 490) in the case definition resulted in a greater prevalence of COPD in women than in men overall, and in the 45 to 64 year age group.

CONCLUSION

Prevalence and mortality measured with administrative health data do not show evidence of relative increase in the prevalence of COPD for women in British Columbia. However, further analysis of ICD-9 490 may identify an early 'at-risk' group, specifically in women.

Validity of gout diagnoses in administrative data

OBJECTIVE

To determine the utility of using administrative data for epidemiologic studies of gout by examining the validity of gout diagnoses in claims data.

METHODS

From a population of approximately 800,000 members from 4 managed care plans, we identified patients who had at least 2 ambulatory claims for a diagnosis of gout between January 1, 1999 and December 31, 2003. From this group, a random sample of 200 patients was chosen for medical record review. Trained medical record reviewers abstracted gout-related clinical, laboratory, and radiologic data from the medical records. Two rheumatologists independently evaluated the abstracted information and assessed whether the gout diagnosis was probable/definite or unlikely/insufficient information. Discordant physician ratings were adjudicated by consensus. Based on record reviews, patients were also classified according to the American College of Rheumatology (ACR), Rome, and New York gout criteria and these results were compared with the physician global assessments.

RESULTS

There were 121 patients rated as having probable/definite gout by physician consensus, leading to a positive predictive value of >or=2 coded diagnoses of gout of 61% (95% confidence interval 53-67). There was low concordance between physician assessments and established gout criteria including ACR, Rome, and New York criteria (kappa = 0.17, 0.16, and 0.20, respectively).

CONCLUSION

Use of administrative data alone in epidemiologic and health services research on gout may lead to misclassification. Medical record reviews for validation of claims data may provide an inadequate gold standard to confirm gout diagnoses.

Diagnostic variability for schizophrenia and major depression in a large public mental health care system dataset

Administrative datasets can provide information about mental health treatment in real world settings; however, an important limitation in using these datasets is the uncertainty regarding psychiatric diagnosis. To better understand the psychiatric diagnoses, we investigated the diagnostic variability of schizophrenia and major depression in a large public mental health system. Using schizophrenia and major depression as the two comparison diagnoses, we compared the variability of diagnoses assigned to patients with one recorded diagnosis of schizophrenia or major depression. In addition, for both of these diagnoses, the diagnostic variability was compared across seven types of treatment settings. Statistical analyses were conducted using t tests for continuous data and chi-square tests for categorical data. We found that schizophrenia had greater diagnostic variability than major depression (31% vs. 43%). For both schizophrenia and major depression, variability was significantly higher in jail and the emergency psychiatric unit than in inpatient or outpatient settings. These findings demonstrate that the variability of psychiatric diagnoses recorded in the administrative dataset of a large public mental health system varies by diagnosis and by treatment setting. Further research is needed to clarify the relationship between psychiatric diagnosis, diagnostic variability and treatment setting.

Medicaid expenditures for children with autistic spectrum disorders: 1994 to 1999

This study used data from 1994 to 1999 from one large county in Pennsylvania to estimate the Medicaid expenditures of children diagnosed with autism spectrum disorders (ASD) and to compare these expenditures with those of other Medicaid-eligible children. On average, children diagnosed with ASD had expenditures 10 times those of other children. Differences in expenditures were driven in large part by inpatient psychiatric care. Further research is required to determine whether hospitalized children could be served in less restrictive and less expensive settings. Lack of differences in ambulatory care expenditures suggests that children with ASD are not receiving additional primary care services that would be indicative of appropriately coordinated services as suggested by the medical home model.

Data quality in an information-rich environment: Canada as an example

This review evaluates the quality of available administrative data in the Canadian provinces, emphasizing the information needed to create integrated systems. We explicitly compare approaches to quality measurement, indicating where record linkage can and cannot substitute for more expensive record re-abstraction. Forty-nine original studies evaluating Canadian administrative data (registries, hospital abstracts, physician claims, and prescription drugs) are summarized in a structured manner. Registries, hospital abstracts, and physician files appear to be generally of satisfactory quality, though much work remains to be done. Data quality did not vary systematically among provinces. Primary data collection to check place of residence and longitudinal follow-up in provincial registries is needed. Promising initial checks of pharmaceutical data should be expanded. Because record linkage studies were ''conservative'' in reporting reliability, the reduction of time-consuming record re-abstraction appears feasible in many cases. Finally, expanding the scope of administrative data to study health, as well as health care, seems possible for some chronic conditions. The research potential of the information-rich environments being created highlights the importance of data quality.

Factors Associated with Celecoxib and Rofecoxib Utilization

BACKGROUND:

The cyclooxygenase-2 (COX-2) selective nonsteroidal antiinflammatory drugs (NSAIDs) celecoxib and rofecoxib (before its removal) are marketed as having fewer gastrointestinal (GI)-related complications than nonselective NSAIDs. However, adverse reaction data suggest that the use of COX-2 selective NSAIDs is associated with clinically significant GI events.

OBJECTIVE:

To assess whether patients receiving celecoxib and rofecoxib have a greater underlying disease burden than patients prescribed nonselective NSAIDs.

METHODS:

The study population consisted of members of 11 health plans, aged >34 years, with a pharmacy claim for celecoxib or rofecoxib or a nonselective NSAID dispensed between February 1, 1999, and July 31, 2001, who had been continuously enrolled for >364 days before the dispensing date. Celecoxib and rofecoxib patients were randomly selected without replacement from a pool of eligible users in each of the 30 months. Nonselective NSAID users were randomly chosen without replacement within each month on a 2:1 ratio to cases; they could be chosen in more than one month. Univariate analyses comparing 9000 cases and 18 000 controls were performed, followed by a multiple logistic regression analysis conditioned on time.

RESULTS:

Increasing age, treatment by a rheumatologist or an orthopedic specialist, treatment with a high number of different medications in the past year, treatment with oral corticosteroids in the past year, and having had a previous GI bleed increased the likelihood of receiving celecoxib or rofecoxib, whereas treatment with a high number of nonselective NSAID prescriptions in the past year decreased it. Treatment with a high number of different medications was a predictor of increased prevalence of underlying diabetes mellitus and cardiovascular disease.

CONCLUSIONS:

Patients having a greater underlying disease burden were more likely to receive COX-2 selective NSAIDs than nonselective ones. Paradoxically, patients at higher risk for cardiovascular disease were channeled toward treatment with COX-2 selective NSAIDs, many of which may confer an increased risk of acute myocardial infarction and other adverse cardiovascular outcomes.

Concordance on the recording of cancer in the Saskatchewan Cancer Agency Registry, hospital charts and death registrations

Accurate and complete registries are an important source of knowledge about cancer. The concordance of the recording of neoplasms in the Saskatchewan cancer registry with that in hospital charts and death registrations was evaluated for 368 patients. The agreement between registry and hospital charts or death registrations was excellent (kappa: 0.93; 95% confidence interval: 0.89, 0.97), with 91.3% of those with cancer having the same neoplasm recorded in their chart or death registration as in the registry. There was only one patient whose hospital chart indicated cancer who was not in the registry and one apparent major discrepancy relating to the cancer site, which was due to the recording of the primary site in the registry and a secondary in the hospital chart. Although based on a relatively small number of patients, these results suggest a high degree of consistency between cancer registry, hospital charts and death registrations in Saskatchewan.

Evaluating the predictive value of osteoarthritis diagnoses in an administrative database

OBJECTIVE

To assess the positive and negative predictive values of osteoarthritis (OA) diagnoses contained in an administrative database.

METHODS

We identified all members (> or =18 years of age) of a Massachusetts health maintenance organization with documentation of at least one health care encounter associated with an OA diagnosis during the period 1994-1996. From this population, we randomly selected 350 subjects. In addition, we randomly selected 250 enrollees (proportionally by the age and sex of the 350 subjects) who did not have a health care encounter associated with an OA diagnosis. Trained nurse reviewers abstracted OA-related clinical, laboratory, and radiologic data from the medical records of both study groups (all but 1 chart was available for review). Pairs of physician reviewers evaluated the abstracted information for both groups of subjects and rated the evidence for the presence of OA according to 3 levels: definite, possible, and unlikely.

RESULTS

Among the group of patients with an administrative diagnosis of OA, 215 (62%) were rated as having definite OA, 36 (10%) possible OA, and 98 (28%) unlikely OA, according to information contained in the medical record. The positive predictive value of an OA diagnosis was 62%. In those without an administrative OA diagnosis, 44 (18%) were assigned a rating of definite OA. The negative predictive value of the absence of an administrative OA diagnosis was 78%.

CONCLUSION

Use of administrative data in epidemiologic and health services research on OA may lead to both case misclassification and under ascertainment.