The validity of administrative data to identify hip fractures is high--a systematic review

The Incidence of Hip Fractures in Long-Term Care Homes in Saskatchewan from 2008 to 2012: an Analysis of Provincial Administrative Databases

Background

Hip fractures (HFs) represent an important cause of morbidity and mortality among adults in long-term care (LTC), but lack of detailed epidemiological data poses challenges to intervention planning. We aimed to determine the incidence of HFs among permanent LTC residents in Saskatchewan between 2008 and 2012, using linked, provincial administrative health databases, exploring associations between outcomes and basic individual and institutional characteristics.

Methods

We utilized the Ministry of Health databases to select HF cases based on ICD 10 diagnoses fracture of head and neck of femur, pertrochanteric fracture and subtrochanteric fracture of femur. HF incidence rates in LTC were compared to older adults in the general population.

Results

LTC residents were more likely to be female overall (65.5%), although this varied by age, with only 46.6% female in those under 65, but 77% female among those 90 years and older. Mean age of residents was highest in rural centres (85.2 yrs) and lowest in medium–large centres (81.0 yrs). Of 6,230 cases of HFs in the province during the study period, 2,743 (44%) were in the LTC cohort. Incidence rates per 1,000 person years increased with age and were higher in the LTC group (F = 68.6, M = 49.8) than the overall population (F = 1.62, M = 0.73). Rates of HFs in the province and in LTC were higher in females than males in all age groups, except for the youngest (< 65 years), where males had higher rates, and the oldest category (90+) where rates were similar. Women 90+ years in larger LTC had significantly higher (p = .035) HF rates than those in smaller LTC, and also had significantly (p = .001) higher rates in medium-large compared to smaller population centres. However, after age standardization to the overall SK population, it was apparent that the larger LTC facilities and the medium-large population centres had overall lower HF rates than the small and medium LTC facilities and the small urban and rural PCs, respectively. One health region had particularly high rates, even when accounting for age and sex composition.

Conclusion

Both HF numbers and incidence rates were higher in LTC compared to the overall population, with higher rates in older women, small to medium size LTC, and particular health regions. Our data suggest the need for further exploration of potentially remediable factors for HFs in smaller LTCs, and for targeting specific facilities and regions with outlying HF rates.

Validity of routinely collected data in identifying hip fractures at a major tertiary hospital in Australia

OBJECTIVES

To examine the validity of routinely collected data in identifying hip fractures (HFs) and to identify factors associated with incorrect coding.

METHOD

In a prospective cohort study between January 2014 and June 2016, HFs were identified using physician diagnosis and diagnostic imaging and were recorded in a Registry. Records of HFs in the health information exchange (HIE) were identified using International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification/Australian Classification of Health Interventions/Australian Coding Standards codes. New HFs were estimated by episode of care, hospital admission and with an algorithm. Data from the HIE and the Registry were compared.

RESULTS

The number of HFs as the principal diagnosis (PD) recorded by episode (864) was higher than by admission (743), by algorithm (711) and in the Registry (638). The sensitivity was high for all methods (90-93%) but the positive predictive value was lower for episode (68%) than for admission (80%) or algorithm (81%). The number of HFs with surgery recorded in the PD by episode (639), algorithm (630) and in the Registry (623) was similar but higher than by admission (589). The episode and algorithm methods also had higher sensitivity (91-92%) than the admission method (84%) for HFs with surgery. Factors associated with coding errors included subsequent HF, long hospital stay, intracapsular fracture, younger age, male, HF without surgery and death in hospital.

CONCLUSIONS

When it is not practical to use the algorithm for regular monitoring of HFs, using PD by admission to estimate total HFs and PD by episode in combination with a procedure code to estimate HFs with surgery can produce robust estimations.

Fracture Risk After Bariatric Surgery: Roux-en-Y Gastric Bypass Versus Adjustable Gastric Banding

The long-term consequences of bariatric surgery on fracture risk are unclear but are likely to vary by procedure type. In physiologic studies, Roux-en-Y gastric bypass (RYGB) and adjustable gastric banding (AGB) have differential effects on rates of bone loss. Therefore, our objective was to compare fracture risk in obese adults after RYGB and AGB procedures. Using claims data from a US commercial health plan, we analyzed rates of nonvertebral fractures within a propensity score-matched cohort (n = 15,032) of morbidly obese adults who received either RYGB or AGB surgery between 2005 and 2013. A total of 281 nonvertebral fractures occurred during a mean follow-up time of 2.3 ± 1.9 years. RYGB patients had an increased risk of nonvertebral fracture (hazard ratio [HR] = 1.43, 95% confidence interval [CI] 1.13-1.81) compared with AGB patients. In fracture site-specific analyses, RYGB patients had increased risk of fracture at the hip (HR = 1.54, 95% CI 1.03-2.30) and wrist (HR = 1.45, 95% CI 1.01-2.07). Nonvertebral fracture risk associated with RYGB manifested >2 years after surgery and increased in subsequent years, with the highest risk in the fifth year after surgery (HR = 3.91, 95% CI 1.58-9.64). In summary, RYGB is associated with a 43% increased risk of nonvertebral fracture compared with AGB, with risk increasing >2 years after surgery. Fracture risk should be considered in risk/benefit discussions of bariatric surgery, particularly among patients with high baseline risk of osteoporosis who are deciding between RYGB and AGB procedures. © 2017 American Society for Bone and Mineral Research.

Comparative Safety and Effectiveness of Denosumab Versus Zoledronic Acid in Patients With Osteoporosis: A Cohort Study

Limited head-to-head comparative safety and effectiveness data exist between denosumab and zoledronic acid in real-world healthcare. We aimed to examine the safety and effectiveness of denosumab compared to zoledronic acid with regard to risk of serious infection and cardiovascular disease (CVD) and osteoporotic fracture. We conducted a cohort study using claims data (2009-2013) from a US commercial insurance plan database. We included patients aged ≥50 years who were newly initiated on denosumab or zoledronic acid. The primary outcomes were (1) hospitalization for serious infection; (2) composite CVD endpoint including myocardial infarction, stroke, coronary revascularization, and heart failure; and (3) nonvertebral osteoporotic fracture including hip, wrist, forearm, and pelvic fracture. To control for potential confounders, we used 1:1 propensity score (PS) matching. Cox proportional hazards models compared the risk of serious infection, CVD, and osteoporotic fracture within 365 days after initiation of denosumab versus zoledronic acid. After PS matching, a total of 2467 pairs of denosumab and zoledronic acid initiators were selected with a mean age of 63 years and 96% were female. When compared with zoledronic acid, denosumab was not associated with an increased risk of serious infection (HR 0.81; 95% confidence interval [CI], 0.55 to 1.21) or CVD (HR 1.11; 95% CI, 0.60 to 2.03). Similar results were obtained for each component of CVD. The risk of osteoporotic fracture was also similar between groups (HR 1.21; 95% CI, 0.84 to 1.73). This large population-based cohort study shows that denosumab and zoledronic acid have comparable clinical safety and effectiveness with regard to the risk of serious infection, CVD, and osteoporosis fracture within 365 days after initiation of medications. © 2016 American Society for Bone and Mineral Research.

Gout and the Risk of Non-vertebral Fracture

Prior studies suggest an association between osteoporosis, systemic inflammation, and pro-inflammatory cytokines such as interleukin (IL)-1 and IL-6. Conflicting findings exist on the association between hyperuricemia and osteoporosis. Furthermore, it remains unknown whether gout, a common inflammatory arthritis, affects fracture risk. Using data from a US commercial health plan (2004-2013), we evaluated the risk of non-vertebral fracture (ie, forearm, wrist, hip, and pelvis) in patients with gout versus those without. Gout patients were identified with ≥2 diagnosis codes and ≥1 dispensing for a gout-related drug. Non-gout patients, identified with ≥2 visits coded for any diagnosis and ≥1 dispensing for any prescription drugs, were free of gout diagnosis and received no gout-related drugs. Hip fracture was the secondary outcome. Fractures were identified with a combination of diagnosis and procedure codes. Cox proportional hazards models compared the risk of non-vertebral fracture in gout patients versus non-gout, adjusting for more than 40 risk factors for osteoporotic fracture. Among gout patients with baseline serum uric acid (sUA) measurements available, we assessed the risk of non-vertebral fracture associated with sUA. We identified 73,202 gout and 219,606 non-gout patients, matched on age, sex, and the date of study entry. The mean age was 60 years and 82% were men. Over the mean 2-year follow-up, the incidence rate of non-vertebral fracture per 1,000 person-years was 2.92 in gout and 2.66 in non-gout. The adjusted hazard ratio (HR) was 0.98 (95% confidence interval [CI] 0.85-1.12) for non-vertebral fracture and 0.83 (95% CI 0.65-1.07) for hip fracture in gout versus non-gout. Subgroup analysis (n = 15,079) showed no association between baseline sUA and non-vertebral fracture (HR = 1.03, 95% CI 0.93-1.15), adjusted for age, sex, comorbidity score, and number of any prescription drugs. Gout was not associated with a risk of non-vertebral fracture. Among patients with gout, sUA was not associated with the risk of non-vertebral fracture. © 2016 American Society for Bone and Mineral Research.

Persistence with denosumab and zoledronic acid among older women: a population-based cohort study

Persistence to denosumab or zoledronic acid was increased compared to oral bisphosphonates.

INTRODUCTION

Denosumab and zoledronic acid are alternative therapies to oral bisphosphonates. Few studies have assessed persistence of those agents.

METHODS

Incident users of denosumab and zoledronic acid were identified using healthcare databases of public drug insurance plan of Quebec province, Canada. Patients initiating therapy between October 1, 2008, and June 30, 2013, and aged 50 years and over were eligible. A persistence rate was assessed over a 2-year period. We assess the proportion of patients receiving the second, third, and fourth injections within a specific delay of predicted time of renewal of both agents. The predictors of non-persistence were analyzed using a Cox regression model only among women.

RESULTS

Among 12,689 incident users, 97.2 % were women. Kaplan-Meier analysis showed a slow decline of persistence after initiating zoledronic acid compared to denosumab therapy, dropping to 81.6 and 63.3 % after 1 and 2 years of follow-up using the permissive gaps of 56 days, in contrast to zoledronic acid, where persistence rate still stays at 74.8 % after 2 years of follow-up using the permissive gap of 112 days. The likelihood of non-persistence was significantly higher among new users of denosumab and zoledronic acid among older patients and year of initiation; but depression and diabetes are only predictors of non-persistence among the zoledronic group. Concomitant use of calcium and vitamin D supplements was at low level which may compromise the clinical efficacy.

CONCLUSION

The persistence rate to denosumab and zoledronic acid was higher to the published data of oral bisphosphonates. The second intention of treatment seems to target more severe patients which may more likely to be compliant.

Modification of claims-based measures improves identification of comorbidities in non-elderly women undergoing mastectomy for breast cancer: a retrospective cohort study

Background

Accurate identification of underlying health conditions is important to fully adjust for confounders in studies using insurer claims data. Our objective was to evaluate the ability of four modifications to a standard claims-based measure to estimate the prevalence of select comorbid conditions compared with national prevalence estimates.

Methods

In a cohort of 11,973 privately insured women aged 18–64 years with mastectomy from 1/04–12/11 in the HealthCore Integrated Research Database, we identified diabetes, hypertension, deficiency anemia, smoking, and obesity from inpatient and outpatient claims for the year prior to surgery using four different algorithms. The standard comorbidity measure was compared to revised algorithms which included outpatient medications for diabetes, hypertension and smoking; an expanded timeframe encompassing the mastectomy admission; and an adjusted time interval and number of required outpatient claims. A χ2 test of proportions was used to compare prevalence estimates for 5 conditions in the mastectomy population to national health survey datasets (Behavioral Risk Factor Surveillance System and the National Health and Nutrition Examination Survey). Medical record review was conducted for a sample of women to validate the identification of smoking and obesity.

Results

Compared to the standard claims algorithm, use of the modified algorithms increased prevalence from 4.79 to 6.79 % for diabetes, 14.75 to 24.87 % for hypertension, 4.23 to 6.65 % for deficiency anemia, 1.78 to 12.87 % for smoking, and 1.14 to 6.31 % for obesity. The revised estimates were more similar, but not statistically equivalent, to nationally reported prevalence estimates. Medical record review revealed low sensitivity (17.86 %) to capture obesity in the claims, moderate negative predictive value (NPV, 71.78 %) and high specificity (99.15 %) and positive predictive value (PPV, 90.91 %); the claims algorithm for current smoking had relatively low sensitivity (62.50 %) and PPV (50.00 %), but high specificity (92.19 %) and NPV (95.16 %).

Conclusions

Modifications to a standard comorbidity measure resulted in prevalence estimates that were closer to expected estimates for non-elderly women than the standard measure. Adjustment of the standard claims algorithm to identify underlying comorbid conditions should be considered depending on the specific conditions and the patient population studied.

Electronic supplementary material

The online version of this article (doi:10.1186/s12913-016-1636-7) contains supplementary material, which is available to authorized users.

Quality of administrative health databases in Canada: A scoping review

OBJECTIVE

Administrative health databases are increasingly used to conduct population-based health research and surveillance; this has resulted in a corresponding growth in studies about their quality. Our objective was to describe the characteristics of published Canadian studies about administrative health database quality.

METHODS

PubMed, Scopus, and Google Advanced were searched, along with websites of relevant organizations. English-language studies that evaluated the quality of one or more Canadian administrative health databases between 2004 and 2014 were selected for inclusion. Extracted information included data quality concepts and measures, year and type of publication, type of database, and geographic origin.

SYNTHESIS

More than 3,000 publications were identified fromthe search. Twelve reports and 144 peer-reviewed papers were included. The majority (53.5%) of peer-review publications used databases from Ontario and Alberta, while 67% of the non-peer-review publications used data from multiple provinces/ territories. Almost all peer-reviewed papers (97.2%) were validation studies. Hospital discharge abstracts and physician billing claims were the most frequently validated databases. Approximately half of the publications (53.0%) validated case definitions and 37.7% focused on a chronic physical health condition.

CONCLUSION

Gaps in the Canadian administrative data quality literature include a limited number of studies evaluating data from the Maritimes and across multiple jurisdictions, newer data sources, validating methods for identifying individuals with mental illness, and assessing the completeness and serviceability of the data. Data quality studies can aid researchers to understand the strengths and limitations of the data.

Variation between Hospitals with Regard to Diagnostic Practice, Coding Accuracy, and Case-Mix. A Retrospective Validation Study of Administrative Data versus Medical Records for Estimating 30-Day Mortality after Hip Fracture

BACKGROUND

The purpose of this study was to assess the validity of patient administrative data (PAS) for calculating 30-day mortality after hip fracture as a quality indicator, by a retrospective study of medical records.

METHODS

We used PAS data from all Norwegian hospitals (2005-2009), merged with vital status from the National Registry, to calculate 30-day case-mix adjusted mortality for each hospital (n = 51). We used stratified sampling to establish a representative sample of both hospitals and cases. The hospitals were stratified according to high, low and medium mortality of which 4, 3, and 5 hospitals were sampled, respectively. Within hospitals, cases were sampled stratified according to year of admission, age, length of stay, and vital 30-day status (alive/dead). The final study sample included 1043 cases from 11 hospitals. Clinical information was abstracted from the medical records. Diagnostic and clinical information from the medical records and PAS were used to define definite and probable hip fracture. We used logistic regression analysis in order to estimate systematic between-hospital variation in unmeasured confounding. Finally, to study the consequences of unmeasured confounding for identifying mortality outlier hospitals, a sensitivity analysis was performed.

RESULTS

The estimated overall positive predictive value was 95.9% for definite and 99.7% for definite or probable hip fracture, with no statistically significant differences between hospitals. The standard deviation of the additional, systematic hospital bias in mortality estimates was 0.044 on the logistic scale. The effect of unmeasured confounding on outlier detection was small to moderate, noticeable only for large hospital volumes.

CONCLUSIONS

This study showed that PAS data are adequate for identifying cases of hip fracture, and the effect of unmeasured case mix variation was small. In conclusion, PAS data are adequate for calculating 30-day mortality after hip-fracture as a quality indicator in Norway.

Fracture Incidence in Adult Kidney Transplant Recipients

BACKGROUND

It remains uncertain whether kidney transplant recipients are a high-risk group for fracture.

METHODS

We conducted a cohort study using Ontario, Canada health care databases to estimate the 3-, 5- and 10-year cumulative incidence of nonvertebral fracture (proximal humerus, forearm, hip) in adult kidney transplant recipients between 1994 and 2009, stratifying by sex and age (<50 versus ≥50 years) at transplant. We also assessed the 3-year cumulative incidence of all fracture locations (excluding skull, toes, and fingers) and falls, 10-year cumulative incidence of hip fracture alone, and nonvertebral fracture incidence in recipients compared to nontransplant reference groups matched on age, sex, and cohort entry year. We studied 4821 recipients (median age, 50 years).

RESULTS

Among the age and sex strata, female recipients aged 50 years or older had the highest 3-year cumulative incidence of nonvertebral fracture (3.1%; 95% confidence interval [95% CI], 2.1-4.4%). Recipients had a higher 3-year cumulative incidence of nonvertebral fracture (1.6%; 95% CI, 1.3-2.0%) compared to the general population with no previous nonvertebral fracture (0.5%; 95% CI, 0.4-0.6%; P < 0.0001) and nondialysis chronic kidney disease (1.1%; 95% CI, 0.9-1.2%; P = 0.03), but a lower fracture incidence than the general population with a previous nonvertebral fracture (2.3%; 95% CI, 1.9-2.8%; P = 0.007). The 10-year cumulative incidence of hip fracture in all recipients was 1.7% (≥3% defined as high risk in clinical guidelines).

CONCLUSIONS

Kidney transplant recipients may have a lower fracture risk than previously suggested in the literature. Results inform our understanding of fracture incidence after kidney transplantation and how it compares to nontransplant populations.

The mortality and direct medical costs of osteoporotic fractures among postmenopausal women in Taiwan

This study estimated the fracture-related mortality and direct medical costs among postmenopausal women in Taiwan by fracture types and age groups by utilizing a nationwide population-based database. Results demonstrated that hip fractures constituted the most severe and expensive complication of osteoporosis across fracture sites.

INTRODUCTION

The aims of the study were to evaluate the risk of death and direct medical costs associated with osteoporotic fractures by fracture types and age groups among postmenopausal women in Taiwan.

METHODS

This nationwide, population-based study was based on data from the National Health Insurance Research Database in Taiwan. Female patients aged 50 years and older in the fracture case cohort were matched in 1:1 ratio with randomly selected subjects in the reference control cohort by age, income-related insurance amount, urbanization level, and the Charlson comorbidity index. There were two main outcome measures of the study: age-differentiated mortality and direct medical costs in the first and subsequent years after osteoporotic fracture events among postmenopausal women. The bootstrap method by resampling with replacement was conducted to generate descriptive statistics of mortality and direct medical costs of the case and control cohorts. Student's t tests were then performed to compare mortality and costs between the two cohorts.

RESULTS

A total of 155,466 postmenopausal women in the database met the inclusion criteria for the fracture case cohort, including 22,791 hip fractures, 72,292 vertebral fractures, 15,621 upper end humerus (closed) fractures, 36,774 wrist fractures, and 7,988 multiple fractures. Analytical results demonstrated that patients experiencing osteoporotic fractures were at considerable excess risk of death and incurred substantially higher treatment costs, notably for hip fractures. Furthermore, results also revealed that the risk of mortality increased with advancing age across the spectrum of fracture sites.

CONCLUSIONS

The present study confirmed an excess mortality and higher direct medical costs associated with osteoporotic fractures. Moreover, hip fractures constituted the most severe and expensive complication of osteoporosis among fracture types.

Validity of Diagnostic Codes for Acute Stroke in Administrative Databases: A Systematic Review

OBJECTIVE

To conduct a systematic review of studies reporting on the validity of International Classification of Diseases (ICD) codes for identifying stroke in administrative data.

METHODS

MEDLINE and EMBASE were searched (inception to February 2015) for studies: (a) Using administrative data to identify stroke; or (b) Evaluating the validity of stroke codes in administrative data; and (c) Reporting validation statistics (sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), or Kappa scores) for stroke, or data sufficient for their calculation. Additional articles were located by hand search (up to February 2015) of original papers. Studies solely evaluating codes for transient ischaemic attack were excluded. Data were extracted by two independent reviewers; article quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies tool.

RESULTS

Seventy-seven studies published from 1976-2015 were included. The sensitivity of ICD-9 430-438/ICD-10 I60-I69 for any cerebrovascular disease was ≥ 82% in most [≥ 50%] studies, and specificity and NPV were both ≥ 95%. The PPV of these codes for any cerebrovascular disease was ≥ 81% in most studies, while the PPV specifically for acute stroke was ≤ 68%. In at least 50% of studies, PPVs were ≥ 93% for subarachnoid haemorrhage (ICD-9 430/ICD-10 I60), 89% for intracerebral haemorrhage (ICD-9 431/ICD-10 I61), and 82% for ischaemic stroke (ICD-9 434/ICD-10 I63 or ICD-9 434&436). For in-hospital deaths, sensitivity was 55%. For cerebrovascular disease or acute stroke as a cause-of-death on death certificates, sensitivity was ≤ 71% in most studies while PPV was ≥ 87%.

CONCLUSIONS

While most cases of prevalent cerebrovascular disease can be detected using 430-438/I60-I69 collectively, acute stroke must be defined using more specific codes. Most in-hospital deaths and death certificates with stroke as a cause-of-death correspond to true stroke deaths. Linking vital statistics and hospitalization data may improve the ascertainment of fatal stroke.

Is routine hospital episode data sufficient for identifying individuals with chronic kidney disease? A comparison study with laboratory data

Internationally, investment in the availability of routine health care data for improving health, health surveillance and health care is increasing. We assessed the validity of hospital episode data for identifying individuals with chronic kidney disease compared to biochemistry data in a large population-based cohort, the Grampian Laboratory Outcomes, Morbidity and Mortality Study-II (n = 70,435). Grampian Laboratory Outcomes, Morbidity and Mortality Study-II links hospital episode data to biochemistry data for all adults in a health region with impaired kidney function and random samples of individuals with normal and unmeasured kidney function in 2003. We compared identification of individuals with chronic kidney disease by hospital episode data (based on International Classification of Diseases-10 codes) to the reference standard of biochemistry data (at least two estimated glomerular filtration rates <60 mL/min/1.73 m(2) at least 90 days apart). Hospital episode data, compared to biochemistry data, identified a lower prevalence of chronic kidney disease and had low sensitivity (<10%) but high specificity (>97%). Using routine health care data from multiple sources offers the best opportunity to identify individuals with chronic kidney disease.

Current and future burden of incident hip fractures in New South Wales, Australia

Population ageing presents significant challenges for many developed nations. Accurately forecasting the likely future burden of age-related medical conditions, such as hip fracture, is critical. In this study, we present estimates of the current and future burden of hip fracture in NSW, Australia, providing crucial information for future health care planning.

PURPOSE

The aims of this study were to investigate the burden of hip fracture in Australia's largest state, New South Wales (NSW), and to build a prediction model to forecast the likely future burden of hip fracture from 2016 to 2036 in persons aged 50 years or more.

METHODS

A retrospective population-based cohort study was conducted using NSW hospitalisation data. Standardised incident hip fracture rates and hip fracture-related acute care length of stay and costs were estimated. Predictive negative binomial regression modelling using age, gender and local health district and year covariates together with projected NSW populations was applied to forecast future hip fractures.

RESULTS

Total incident hip fractures increased 8.8 % over a 12-year period from 2000/2001 to 2011/2012 despite declining age-standardised rates. Estimates of acute care length of stay for the treatment of hip fracture ranged from 10 to 15 days and acute care costs ranged between 21 and 29,000 Australian dollars per fracture. By 2036, incident hip fractures are projected to rise by 35.2 %, assuming a continued decline in the rate of hip fracture or by 107.5 % if the current decline in the rate does not continue. Acute care length of stay and costs are each predicted to rise between 37.1 and 110.4 % by 2036.

CONCLUSION

An ageing population and changing demographics will continue to drive the increasing burden of incident hip fractures in NSW and Australia in the foreseeable future. These anticipated changes provide important information for the planning and management of future hip fracture care.

Validity of heart failure diagnoses in administrative databases: a systematic review and meta-analysis

Objective

Heart failure (HF) is an important covariate and outcome in studies of elderly populations and cardiovascular disease cohorts, among others. Administrative data is increasingly being used for long-term clinical research in these populations. We aimed to conduct the first systematic review and meta-analysis of studies reporting on the validity of diagnostic codes for identifying HF in administrative data.

Methods

MEDLINE and EMBASE were searched (inception to November 2010) for studies: (a) Using administrative data to identify HF; or (b) Evaluating the validity of HF codes in administrative data; and (c) Reporting validation statistics (sensitivity, specificity, positive predictive value [PPV], negative predictive value, or Kappa scores) for HF, or data sufficient for their calculation. Additional articles were located by hand search (up to February 2011) of original papers. Data were extracted by two independent reviewers; article quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies tool. Using a random-effects model, pooled sensitivity and specificity values were produced, along with estimates of the positive (LR+) and negative (LR−) likelihood ratios, and diagnostic odds ratios (DOR = LR+/LR−) of HF codes.

Results

Nineteen studies published from1999–2009 were included in the qualitative review. Specificity was ≥95% in all studies and PPV was ≥87% in the majority, but sensitivity was lower (≥69% in ≥50% of studies). In a meta-analysis of the 11 studies reporting sensitivity and specificity values, the pooled sensitivity was 75.3% (95% CI: 74.7–75.9) and specificity was 96.8% (95% CI: 96.8–96.9). The pooled LR+ was 51.9 (20.5–131.6), the LR− was 0.27 (0.20–0.37), and the DOR was 186.5 (96.8–359.2).

Conclusions

While most HF diagnoses in administrative databases do correspond to true HF cases, about one-quarter of HF cases are not captured. The use of broader search parameters, along with laboratory and prescription medication data, may help identify more cases.

Validity of myocardial infarction diagnoses in administrative databases: a systematic review

BACKGROUND

Though administrative databases are increasingly being used for research related to myocardial infarction (MI), the validity of MI diagnoses in these databases has never been synthesized on a large scale.

OBJECTIVE

To conduct the first systematic review of studies reporting on the validity of diagnostic codes for identifying MI in administrative data.

METHODS

MEDLINE and EMBASE were searched (inception to November 2010) for studies: (a) Using administrative data to identify MI; or (b) Evaluating the validity of MI codes in administrative data; and (c) Reporting validation statistics (sensitivity, specificity, positive predictive value (PPV), negative predictive value, or Kappa scores) for MI, or data sufficient for their calculation. Additonal articles were located by handsearch (up to February 2011) of original papers. Data were extracted by two independent reviewers; article quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies tool.

RESULTS

Thirty studies published from 1984-2010 were included; most assessed codes from the International Classification of Diseases (ICD)-9th revision. Sensitivity and specificity of hospitalization data for identifying MI in most [≥50%] studies was ≥86%, and PPV in most studies was ≥93%. The PPV was higher in the more-recent studies, and lower when criteria that do not incorporate cardiac troponin levels (such as the MONICA) were employed as the gold standard. MI as a cause-of-death on death certificates also demonstrated lower accuracy, with maximum PPV of 60% (for definite MI).

CONCLUSIONS

Hospitalization data has higher validity and hence can be used to identify MI, but the accuracy of MI as a cause-of-death on death certificates is suboptimal, and more studies are needed on the validity of ICD-10 codes. When using administrative data for research purposes, authors should recognize these factors and avoid using vital statistics data if hospitalization data is not available to confirm deaths from MI.