Abstract
Objectives. To assess the validity of the diagnoses of atrial fibrillation (AF) and atrial flutter (AFL) for men and women recorded in the Danish National Patient Registry, and to assess the relative distribution of AF and AFL. Design. Review of medical records for incident cases of AF and/or AFL in the Diet, Cancer, and Health cohort study. Participants were enrolled in 1993–97 with 13.6 years of follow-up until 30 December, 2009. Results. The positive predictive value of the combined diagnosis of AF and/or AFL was 92.6% (95% CI 88.8%; 95.2%) with no significant difference between sexes (men 93.7% (133/142), women 90.8% (129/142)). The proportion of AFL either alone or in combination with AF was significantly higher in men than in women (13.5% (18/133) vs. 5.4% (7/129), p =0.03). The positive predictive value of the specified diagnosis of AFL was 57.5% for men (46/80) and 29.6% for women (8/27). Conclusions. This study shows that the validity of the diagnosis of AF and/or AFL is high and may be used for registry-based studies. A specified diagnosis of AFL was rarely used and was not reliable to distinguish between cases of AF and AFL.
Introduction
Hospital discharge registries are cost-effective sources of information on incidence and prevalence of specific diagnoses in a population and are also valuable for cross-linking with cohort studies. Atrial fibrillation (AF) is a common heart rhythm disorder, and the prevalence increases with age from less than 0.5% at age 40–50 years to 5–15% at age 80 (Citation1–5). Medical costs related to AF are enormous and rising as the number of patients with AF is increasing, partly explained by an increasing number of elderly in the population (Citation5).
Traditionally, many observational studies such as the Framingham Heart Study (Citation6) have categorised AF as AF and/or atrial flutter (AFL) acknowledging the clinical interrelationship between these two diagnoses (Citation7). However, with new treatment options in the era of radiofrequency ablation, it may be important to distinguish between the two conditions.
Previous studies have reported a high validity of the combined diagnosis AF and/or AFL with positive predictive values (PPV) above 96% (Citation8–11). As the validity of hospital diagnoses may vary between regions, hospitals, and departments, the present study was initiated to estimate the PPV in a large cohort. Increasing attention has been devoted to the differences in heart disease between men and women in recent years. The aims of the present study were to assess the validity of the diagnoses of AF and AFL for both men and women in the Danish National Patient Registry, and to assess the relative distribution of AF and AFL.
Methods
Subjects
Diet, Cancer, and Health is a cohort study initiated with the primary objective to examine the etiological role of diet in the development of cancer, as described in detail previously (Citation12). Briefly, 57,053 persons aged 50–64 years were enrolled between December 1993 and May 1997. Eligible participants had to be born in Denmark and not registered with a cancer diagnosis in the Danish Cancer Registry at the time of invitation. Participants registered with AF or AFL before inclusion were excluded.
Follow-up was done using information from the Danish National Patient Registry via the personal identification number that uniquely identifies every Danish citizen. In this registry, discharge diagnoses from in-hospital patients have been registered since 1977, and since 1995 diagnoses from emergency rooms and out-patient visits have been recorded as well. The diagnoses were recorded using the 8th International Classification of Diseases (ICD-8) until the end of 1993 (427.93 AF and 427.94 AFL in the Danish version, which is equivalent to 427.4 AF or AFL in the international version). After 1 January, 1994, the ICD-10 classification was used with the diagnosis of AF and/or AFL (I.48) and in the Danish version with the possibility of specifying AFL (I.48.9A) and AF (I.48.9B). Last date for follow-up was 30 December, 2009.
Validation
To assess the validity of the diagnosis AF and/or AFL, hospital medical records were reviewed in a random sample of cases identified in the National Patient Registry, stratified by sex to include 150 men and 150 women. Sampling was performed using the statistical software Stata 11 (StataCorp, USA). Electrocardiograms (ECG) or other heart rhythm documentation was scanned anonymously using a sample identification number and reviewed by two independent reviewers, an MD with cardiologic experience (TR) and the head of electrophysiology at the department (SR). Discrepancies were solved in agreement or by final decision by a second electrophysiologist. If sufficient documentation was not found for the registered event, all accessible medical records were reviewed. AF was defined according to the ESC guidelines as absence of consistent P-waves, fibrillatory waves and irregular R-R intervals (Citation13). AFL was defined as presence of flutter waves (Citation13). AF/AFL was categorised as “confirmed” if relevant documentation (ECG or heart rhythm monitoring) could be provided. Patients treated with radiofrequency ablation for AF or AFL or participating in a randomized controlled trial for AF were also defined as “confirmed.” If a relevant ECG or other heart rhythm documentation could not be retrieved, findings were classified as “probable” if this was determined to be appropriate based on the available information in medical records. Additional information on the study population was also recorded using a pre-defined list, including information on treatment, predisposing factors, and whether the patient was known to have AF or AFL before admission (). Every single case of the specified diagnosis AFL (I.48.9A) was reviewed as described above.
Table I. Patient characteristics at time of first registration of AF and/or AFL.
Statistics
Statistical analysis was done using Stata 11 (StataCorp, USA). PPV was calculated as total of confirmed cases and probable cases divided by the number of medical records reviewed (missing records not counted). Confidence intervals (CI, 95%) were calculated using logit transformation. As the validity may have changed during the history of follow-up, fractional polynomial models (Citation14) were used to model simple smooth non-linear time trends in PPV over calendar time. P-values were calculated using Fisher's exact test, and a two-tailed p-value <0.05 was considered statistically significant.
Results
In the Diet, Cancer, and Health Cohort, 80,996 men and 79,729 women were invited to participate of which 57,053 persons (27,178 men and 29,875 women) accepted (). Forty-two subjects were excluded because of missing questionnaire information. Participants registered with AF or AFL before inclusion were excluded (n =456). Subsequently, a total of 3425 incident cases of AF and/or AFL were registered (2181 men (8.0%), 1244 women (4.2%)) in the National Patient Registry during a median follow-up of 13.6 years).
Figure 1. Flowchart of participants in the Diet, Cancer, and Health cohort study and incident diagnosis of atrial fibrillation (AF) and/or atrial flutter (AFL) recorded in the Danish National Patient Registry. A random sample of 150 men and 150 women of incident cases was selected for review of hospital medical records.
In total, 284 (94.6%) of the randomly selected 300 medical records could be retrieved (142 of 150 for each sex, ). Medical records were obtained for the incident events recorded at 76 departments and 8 emergency rooms in 29 hospitals. Reasons for missing records were—no response from the clinical departments (n =4), hospital or department closed (n =3), or hospital records not possible to retrieve (n =9). AF and/or AFL was confirmed in 229 cases, whereas 33 were categorized as “probable AF” (n =29) or “probable AFL” (n =4). Reasons for missing heart rhythm documentation were: diagnosed at a private cardiologist (n =5), at a general practitioner (n =1), or at another hospital and then transferred (n =2); ECG not available in paper format in the medical records (n =17); ECG not scanned to an electronic patient system (n =5); and finally some patients were known to have AF before admission (n =3). In the following, combined results for “confirmed” and “probable” diagnoses are reported.
Table II. Validation of the diagnosis of AF and/or AFL (I.48) in the Danish National Patient Registry.
The PPV of the combined diagnosis AF and/or AFL (I.48) was 93.7% (133/142) for men and 90.8% (129/142) for women with no significant difference between sexes (PPV 2.8%, CI − 3.4%; 9.0%). For all cases in the cohort, the PPV was 92.6% (CI 88.8%; 95.2%), calculated weight-adjusted according to the higher proportion of cases in men compared to women. Stratification by in-hospital and out-patient visits as well as primary or secondary diagnoses gave similar results. The PPV was significantly lower for the few cases registered in an emergency room (64.7% (11/17), CI 39.9%; 83.5%) than for in-hospital and out-patient visits (94.0% (251/267), CI 90.4%; 96.3%, p <0.01). The PPV for departments of cardiology (93.9% (168/179), CI 89.1%; 96.6%) was not significantly different compared to other departments (89.5% (94/105), CI 81.9%; 94.2%, p =0.25). There was no change in validity over time when tested by fractional polynomial models. The reasons for the 22 incorrect diagnoses were: ECG misinterpretation (n =9); non-documented palpitations (n =5); or diagnoses not described by treating physicians and most likely coded by administrative personnel (n =8). The majority of verified cases had AF (). The proportion of AFL either alone or in combination with AF was significantly higher in men (13.5% (18/133), CI 8.6%; 20.6%) than in women (5.4% (7/129), CI 2.6%; 11.1%, p =0.03).
All incident cases of the specified diagnosis of AFL (I.48.9A) were reviewed (81 men and 27 women, ). Sufficient heart rhythm documentation was missing in 10 cases, of which five were classified as “probable AFL” (AFL described in text) and five as “probable AF” (AF described in text but coded as AFL). The PPV of the diagnosis of AFL was 57.5% in men (46/80, CI 46.2%; 68.0%) and 29.6% in women (8/27, CI 15.3%; 49.4%) for AFL either alone or in combination with AF.
Table III. Validation of the diagnosis of AFL (I.48.9A) in the Danish National Patient Registry.
Discussion
For the combined diagnosis of AF and/or AFL (I.48), we found a PPV of 92.6% with no significant difference between sexes (men 93.7%, women 90.8%). This high validity of the diagnosis in the Danish National Patient Registry indicates that it may be used for registry-based studies. Importantly, this finding is based on a review of a random sample of patients from a number of different types of hospitals and hospital departments. This study indicates that the validity of this diagnosis is high for both sexes, which is important for future registry-based studies comparing different aspects regarding AF between men and women.
The epidemiology of AF has been extensively studied, but there are limited data regarding AFL (Citation5,Citation15). While it is well known that AF is more common in men than in women, the sex difference was even more pronounced for AFL. Among incident cases of AF and/or AFL in our study, 13.5% of men and 5.5% of women had AFL either alone or in combination with AF. When applied to the entire cohort, an estimated 1.0% of men (CI 0.6%; 1.5%) and 0.2% of women (CI 0.1%; 0.4%) developed AFL during 13.6 years of follow-up with corresponding incidence rates of 82/100,000 person-years for men and 16/100,000 person-years for women. However, these rough estimates are based on few cases and should be interpreted cautiously. To our knowledge, the only comparable population study on ECG- verified cases of AFL is from the Marshfield Epidemiological Study Area (Citation15) with a reported 2.5-fold higher incidence of AFL in men than in women.
We also reviewed all incident cases registered with the specified diagnosis AFL (I.48.9A). Our data show that this recorded diagnosis was unreliable to distinguish specified AFL from AF as almost half of diagnoses registered as AFL were actually AF. Compared to the frequency in patients registered as AF and/or AFL (I.48), the specified diagnosis of AFL was rarely used as it included only 54 confirmed cases of AFL of an estimated total of 337 cases in the cohort.
This study has a number of strengths. It validates a random sample of a very large and well-defined cohort with high quality follow-up data, due to continuously updated information on vital status and discharge diagnoses via the personal identification number. Since all Danish hospitals report to the National Patient Registry, a subject diagnosed at any hospital in Denmark is part of this study regardless of primary hospital affiliation. Heart rhythm documentation was evaluated independently by two reviewers for verification of AF or AFL. Medical records reviewed were from a large number of different hospitals and departments and supportive of the generalizability of the results.
There were also limitations to this study. Some medical records did not contain relevant heart rhythm documentation, either because the registration actually referred to a prior diagnosis that had not been recorded in the registry or because of insufficiently stored documentation. In total, 33 of the 262 validated cases (12.6%) were classified as “probable AF” or “probable AFL”, based on available information. However, we do not believe that this significantly affects the estimated overall validity, which to some extent is also supported by the finding that only a minor part (9 of 22) of the incorrectly registered AF and/or AFL diagnoses were due to incorrect interpretation of the ECG.
Only persons registered from a hospital were included. Persons only seen by a general practitioner or not seeking medical attention are not recorded in the National Patient Registry. It can be argued, however, that most first-time cases of AF or AFL found by general practitioners would be referred for cardiac resynchronization or echocardiography and at that time be registered. This is supported by the finding that the majority (82%) of cases were diagnosed less than 6 months before registration (). As regards the external validity, the subjects had chosen to participate in a cohort study and therefore might be a selected group. However, the only requirement for participants was to complete the baseline after which follow-up was done using national registries. Consequently, it is unlikely that the validity of the diagnosis was affected substantially, and it seems reasonable to extrapolate the information on validity to other studies using the National Patient Registry for these age groups.
Recently, a Swedish study (Citation10) from Malmö reported a 97% PPV for the diagnosis AF and/or AF in a random sample of 100 medical records primarily obtained from a single university hospital centre using an electronic ECG database. Our study supplements that study by reviewing a larger sample of Danish medical records for patients from a number of different hospitals and departments, reporting results for both men and women and adding information on the relative proportions of AF and AFL as well as information on the validity of a specified diagnosis of AFL.
The PPV of the combined diagnosis of AF and/or AFL (I.48) in our study was high but somewhat lower than in other studies reporting PPV >96% (Citation8–11,Citation16). Misclassification related to incorrect ECG interpretation accounted for 9 of 284 (3.2%), which is comparable to 4.5% of AF in a report on ECG misinterpretation by Davidenko et al. (Citation17). There was additional misclassification (13/284, 4.6%) related to other factors such as non- documented palpitations and inappropriate coding that did not reflect the information in medical records. To some extent this may reflect the fact that this random sample of 150 men and 150 women represents information from a large number of different hospital departments and also includes emergency room visits. Two invalid diagnoses could be avoided by discarding diagnoses registered in the emergency department on the same day as an admission for an in-hospital stay.
In conclusion, this study confirms earlier reports that the validity of the diagnosis of AF and/or AFL (I.48) is high and may be used for registry-based studies. The specified diagnosis of AFL, however, was rarely used and was not reliable to distinguish between AF and AFL.
Acknowledgement
This study was supported by a research grant from the Danish Council for Strategic Research (grant number 09-066965).
Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
References
- Stewart S, Hart CL, Hole DJ, McMurray JJ. Population prevalence, incidence, and predictors of atrial fibrillation in the Renfrew/Paisley study. Heart. 2001;86:516–21.
- Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV, . Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA. 2001;285:2370–75
- Miyasaka Y, Barnes ME, Gersh BJ, Cha SS, Bailey KR, Abhayaratna WP, . Secular trends in incidence of atrial fibrillation in Olmsted County, Minnesota, 1980 to 2000, and implications on the projections for future prevalence. Circulation. 2006;114:119–25.
- Heeringa J, van der Kuip DA, Hofman A, Kors JA, van Herpen G, Stricker BH, . Prevalence, incidence and lifetime risk of atrial fibrillation: the Rotterdam study. Eur Heart J. 2006;27:949–53.
- Naccarelli GV, Varker H, Lin J, Schulman KL. Increasing prevalence of atrial fibrillation and flutter in the United States. Am J Cardiol. 2009;104:1534–39.
- Lloyd-Jones DM, Wang TJ, Leip EP, Larson MG, Levy D, Vasan RS, . Lifetime risk for development of atrial fibrillation: the Framingham Heart Study. Circulation. 2004; 110:1042–46.
- Waldo AL. The interrelationship between atrial fibrillation and atrial flutter. Prog Cardiovasc Dis. 2005;48:41–56.
- Mozaffarian D, Psaty BM, Rimm EB, Lemaitre RN, Burke GL, Lyles MF, . Fish intake and risk of incident atrial fibrillation. Circulation. 2004;110:368–73.
- Virtanen JK, Mursu J, Voutilainen S, Tuomainen TP. Serum long-chain n-3 polyunsaturated fatty acids and risk of hospital diagnosis of atrial fibrillation in men. Circulation. 2009;120:2315–21.
- Smith JG, Platonov PG, Hedblad B, Engstrom G, Melander O. Atrial fibrillation in the malmo diet and cancer study: a study of occurrence, risk factors and diagnostic validity. Eur J Epidemiol. 2010;25:95–102.
- Frost L, Vestergaard P. Alcohol and risk of atrial fibrillation or flutter: a cohort study. Arch Intern Med. 2004;164:1993–98.
- Tjonneland A, Olsen A, Boll K, Stripp C, Christensen J, Engholm G, . Study design, exposure variables, and socioeconomic determinants of participation in Diet, Cancer and Health: a population-based prospective cohort study of 57,053 men and women in Denmark. Scand J Public Health. 2007;35:432–41.
- Fuster V, Ryden LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen KA, . ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation: full text: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines and the European Society of Cardiology Committee for Practice Guidelines (Writing Committee to Revise the 2001 guidelines for the management of patients with atrial fibrillation) developed in collaboration with the European Heart Rhythm Association and the Heart Rhythm Society. Europace 2006;8:651–745.
- Royston P, Altman DG. Regression using fractional polynomials of continuous covariates: parsimonious parametric modelling. Appl Stat. 1994;43:429–67.
- Granada J, Uribe W, Chyou PH, Maassen K, Vierkant R, Smith PN, . Incidence and predictors of atrial flutter in the general population. J Am Coll Cardiol. 2000;36:2242–46.
- Frost L, Vestergaard P. n-3 Fatty acids consumed from fish and risk of atrial fibrillation or flutter: the Danish Diet, Cancer, and Health Study. Am J Clin Nutr. 2005;81:50–4.
- Davidenko JM, Snyder LS. Causes of errors in the electrocardiographic diagnosis of atrial fibrillation by physicians. J Electrocardiol. 2007;40:450–6.