Cardiovascular diseases

Cardiovascular diseases (algorithm)

LASA filenames:
zh01, zh02, zh03

Contact: Natasja van Schoor

Background
Angina pectoris (AP), myocardial infarction (MI), congestive heart failure (CHF) and cardiac arrhythmia (AR) are highly prevalent cardiac diseases. Peripheral arterial disease (PAD) and cerebrovascular accident (CVA) are major vascular diseases. For these specific diseases, algorithms (decision trees) were developed using data on cardiovascular disorders in the main interview (self-report (SR) of chronic diseases), the medical interview (inspections of medicine bottles) and through the medical records of general practitioners (GP). The aim of making an algorithm is to enhance the reliability of the data on these diagnoses1, 2. In 2018, the existing CVD algorithms were revised (see Cardiovascular diseases before revision here), mainly because of new medical insights. For example, the use of certain medication can be very specific for a disease, such as the use of nitro-glycerine and having angina pectoris. In such cases, the medication use was given more weight in the algorithm. Furthermore, data from the telephone interviews and self-reported surgery were added to the algorithms to reduce missing data. Data on AP, MI and self-reported coronary heart surgery were combined to determine the presence of coronary artery disease.

Data sources
Data files and variables used in the algorithm:

Self-report of chronic diseases:
– Face to face interview: LASA*035;
– Telephone interview with respondent and proxy: LASAC602 and with proxy: LASA*602 (from wave D and on);
– Telephone interview with respondent: LASA*702 (from wave D and on)
Medication (derived by inspection of medication bottles): LASA*352
Medical records of general practitioners: LASA*G01 (datafiles under construction)
Year in which the interview with the respondent took place: t*_dat in LASAZ008
Defining dropouts: *result in LASAZ002

For more information on the questionnaires used to retrieve the data mentioned above, please click on the names in bold underlined (in Dutch).

The algorithms

Main components
The self-report of CVD was retrieved from the main interview. For some of the CVDs, the presence of the disease was asked explicitly (MI, AR, CVA) and for other CVDs, the presence of specific symptoms were asked to determine the SR presence of a CVD (AP, CHF, PAD). The questions considering somatic disease were so called branch-questions, i.e. further questions were only asked in case of a positive answer on the previous questions. For example, if a respondent reported to have no heart disease of myocardial infarction, all subsequent questions on cardiac disease were skipped. In addition to this, data from the telephonic interviews with respondents or proxies were used. Again, some CVDs were asked explicitly in the telephonic interview (CVA), while other questions were too broad to use as a confirmation of specific CVDs. For these questions, however, a negative answer was used to determine the absence of CVDs (AP, MI, CHF, AR, PAD). For example, during the telephonic interview ‘does [the respondent] have a heart disease or myocardial infarction?’ was asked. A positive answer to this question cannot be linked to one specific CVD. A negative answer, however, means that the respondent did not have a SR diagnosis of AP, MI, CHF and AR.
Surgery: self-report of specific types of surgery on the cardiovascular system.
Medication use: the respondents were asked to bring in the medication with packaging that they used in the past two weeks. All the medication was noted. Based on ATC-codes, specific types of medication were used as an indicator of AP, CHF, CAD (syntax available upon request), see Medication use.
GP diagnosis of CVD: In 1994, 2000/2001, 2005/2006, 2010 and 2016 general practitioners were asked to fill in a questionnaire with specific questions on the medical history of respondents.

As mentioned above, not every component of the algorithm is available for every CVD. For an overview, see table 1.

Table 1. Availability of components of the algorithms

na: data not available. ^aIn the telephone interview, only one general question on heart disease was asked (‘do you (or the respondent) have a heart disease?’) and no questions on specific types of heart disease. Therefore, only a negative answer was used in the algorithm.

Combining information on these aspects of CVD and non-response data, respondents are categorized in 6 groups for every CVD: definite / possible / contradictory / no / missing / dropout.

The seven algorithms can be found here:
– Angina pectoris (PDF)
– Myocardial infarction (PDF)
– Coronary artery disease (PDF)
– Cardiac arrhythmia (PDF)
– Cerebrovascular disease (PDF)
– Congestive heart failure (PDF)
– Peripheral artery disease (PDF)
A more detailed description of these algorithms and the decisions that were made during the revision, is available upon request.

The syntax of the algorithms can be found here:
– Syntax first cohort (PDF)
– Syntax second cohort (PDF)
– Syntax third cohort (PDF)

Availability of information per wave

Table 2. Prevalence of CVD (any type of CVD)

Previous use in LASA
Revised algorithm: no publications yet.

Original algorithm:

Penninx BW, Beekman AT, Honig A, Deeg DJ, Schoevers RA, van Eijk JT, Van Tilburg W. Depression and cardiac mortality: results from a community-based longitudinal study. Arch Gen Psychiatry 2001;58:221-7

References

1. Smith, B., Chu, L., Smith, T., Amoroso, P., Boyko, E., & Hooper, T. et al. (2008). Challenges of self-reported medical conditions and electronic medical records among members of a large military cohort. BMC Medical Research Methodology, 8(1). doi: 10.1186/1471-2288-8-37
2. Polubriaginof CG, F., & Pastore G, P. (2016). Comparing Patient-Reported Medical Problems with the Electronic Health Record Problem List. General Medicine: Open Access, 04(03). doi: 10.4172/2327-5146.1000258

Date of last update: April 22, 2020