Chronic diseases

Cardiovascular diseases (algorithm)

Contact: Bianca Schalk, Marijke Bremmer (, Marjolein Visser)

Background
Angina Pectoris (AP), Myocard Infarct (MI), Congestive Heart Failure (CHF) and Cardiac Arrhythmia (AR) are highly prevalent cardiac diseases. Peripheral Arterial Disease (PAD) and CerebroVasculair 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 of chronic diseases), the medical interview (inspections of medicine bottles) and through the medical records of general practitioners. The aim of making an algorithm is to enhance the reliability of the self-report data on these diagnoses.

Sources
Diagnoses of these diseases are confirmed by three data resources:
• Self-reported (symptoms of) (cardio)vascular diseases
LASA(B-C-D)035: chronic diseases
• Medication use during the past two weeks registered by medical interviewers
All prescribed drugs were pooled in specific groups per indication, e.g. anti-hypertensives, diuretics. For detailed information see documentation on: ‘pharmacotherapy for cardiovascular diseases in LASA’ (LASAb552, C352, D352 M. Bremmer).
• Medical records of General Practitioners (GP) (see appendix 1)
1) GP data of 1992/93 (D. Kriegsman and B. Penninx; see article D.Kriegsman(1))). These data are collected in 1992/1993 and therefore have only information until 1993. The GP questionnaire of 1992/93 contains only questions about CVA, PAD and one general question on not further specified cardiac disease (yes/no). Therefore the GP questionnaire of 1992/93 can only be used in the specific algorithms of CVA and PAD. The general question about cardiac disease (yes/no) can only be used in the pooled variable on cardiac diseases.
2) GP data obtained in 1999/2001 (For detailed information see ‘LASACG01: Documentation second General Practitioner collection (2000/01)’) the timeframe was before 1992, 1992 etc. until 2001. In this GP data collection specific questions were asked on when cardiac disease was present, and also ‘did a specialist confirm this disease’. If the GP data and the self-report data were inconclusive (leading to a “possible” diagnosis) the specialist information on the GP questionnaire decided whether the diagnosis was changed into ‘definite’ cardiac disease.
At baseline, each specific cardiovascular disease was confirmed by GP when general practitioner diagnosed this disease before 1992 or in the year of the interview: 1992 or 1993. For the C-cycle and D-cycle the same was done, using the time frame between the years of interviewing 1995/96 and 1998/99. However when interview year was missing at the d-cycle (thus no main interview at the d-cycle) and the disease was diagnosed in 1999 by the GP, we assumed confirmation for this disease by the GP at the d-cycle.

Variable information
These three data resources are combined to make algorithms separate for each cardiovascular disease (AP, MI, CHF, AR, PAD and CVA) which are being described in the next part ‘specific algorithms for each cardiovascular diseases’. We have made a categorical variable for each cardiovascular disease separately (no, possible or definite) and furthermore we have constructed two pooled variables.

Algorithm for each cardiovascular disease
In short, based on the main sources self-report, medication and GP information the algorithm for each specific cardiovascular disease is:
• Definite when at least two of the main sources are ‘yes’ or the specialist confirmation makes the ‘possible’ ‘definite’
• Possible, all other options.
• No, when at least two of the main sources are answered ‘no’.
• Missing value, if all main sources were missing.

Pooled variables: cardiac disease and cardiovascular diseases
All cardiac diseases include the variables MI, CHF, AP and AR and all cardiovascular diseases, adds CVA and PAD. These two variables are categorized in three strata according to the level of certainty of the diagnosis:
• Definite, at least one cardio(vascular) diseases is definite.
• Possible, at least one cardio(vascular) disease is possible, but no definite diagnosis.
• No, no possible nor definite cardiovascular diagnosis.
• Missing value, if all cardio(vascular) diseases were missing.
In sum, we created a new data file ‘LASAZH01.sav for the B-, C and D-cycle containing structured variables of AP, MI, CHF, AR, PAD and CVA apart, cardiac disease and cardiovascular diseases.

Considerations and limitations
Medical records
As mentioned before, at baseline we have used medical record from the GP obtained in 1992/93 as well as the information obtained in 2000/01 pertaining to respondents’ medical status in 1992-1993. Only in the specific algorithms of CVA and PAD the GP information derived in 1992/93 and 2000/01 are used together. When one of the GP data was answered ‘yes’, the overall GP information in the algorithm was ‘yes’. In the other specific algorithms only information of the GP information derived in 2000/01 was used. In addition, the general question about cardiac disease (yes/no) could only be used in the pooled variables of cardiovascular diseases and cardiac diseases. In the algorithm of LASA-C (1995/96) and LASA-D (1998/99) we used solely the latest medical records obtained in 2000/01.

Longitudinal data-cleaning
In general, once a respondent had a cardiovascular disease (s)he is affected by this chronic disease for the rest of his(her) life. Clinically, this holds true for all cardiovascular conditions under study.
The longitudinal data-cleaning was done as follows: if (for instance) AP was definitely present at the B-cycle and according to the algorithm at LASA-C ‘not present’ or ‘possibly present’, the value of the C-cycle was recoded into ‘3’; meaning “definite at the B-cycle and C-cycle AP not or possibly”. The same was done at the D-cycles, where ‘3’ means “definite at the B-cycle or C-cycle and D-cycle AP not or possibly”. For longitudinal analyses of results it is recommended to further recode this value 3 (definite at the B-cycle and C-cycle AP not or possibly; definite at the B-cycle or C-cycle and D-cycle AP not or possibly) into 1 (definite), but the individual researcher should make this decision. This rule was applied to all cardiovascular disease variables, including the pooled variables.

Myocardial infarction and cerebrovascular accidents
In the main interview at all three LASA-cycles the respondents were asked the same questions: have you or have you ever had.... (heart disease/ chest pain in exertion). Only in the case of MI and CVA the questions differ between the B- versus the C-and D-cycle. ‘Have you ever had a MI?’ at baseline and in the C- and D-cycle: ‘Have you had a MI since our last interview?’. The two diseases have a recurrent incidental course. The longitudinal cleaned variables represent the first and the following incidents together. For example, when at baseline no MI was diagnosed, and at three-year follow-up MI was diagnosed, we do not know if the respondent has had one or more MI’s. We have not been able, as yet, to construct one or more variables in which the number and year of each accident is represented separately.

Previous use of cardiovascular data in LASA
Brenda Penninx (2) has published one article using a pooled variable for all cardiac diseases. As we were able to use the GP information on separate cardiac diseases (GP data 2000/01), the algorithms on angina pectoris, congestive heart failure, arrhythmia’s and myocardial infarctions are different from the algorithms by B. Penninx.

Reference List

  1. Kriegsman DM, Penninx BW, van Eijk JT, Boeke AJ, Deeg DJ. Self-reports and general practitioner information on the presence of chronic diseases in community dwelling elderly. A study on the accuracy of patients' self-reports and on determinants of inaccuracy. J Clin Epidemiol 1996;49:1407-17.
  2. 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.


Specific algorithms for each cardiovascular disease:

Angina Pectoris (=AP)

Myocardial infarction (=MI)

Congestive heart failure (=CHF)

Cardiac arrhythmia’s (=AR)

Peripheral arterial disease (=PAD)

Cerebro Vascular Accident (=CVA)

 

Number of cardiac disease at the three LASA cycles before and after longitudinal data-cleaning.

 
Before
After
Cycle B C D B C D
Missing
16 675 891 Idem 675 891
NO
2433 2022 1820   1858 1602
Definite
458 201 212   201 212
Possible
200 209 184   123 96
Definite cardiac disease at LASA B
- - -   250 -
Definite cardiac disease at LASA B or C
- - -   - 306

 

Number of cardiovascular diseases (CVD) at the three LASA cycles before and after longitudinal data-cleaning.

 
Before
After
Cycle B C D B C D
Missing
14 675 890 Idem 675 890
NO
2195 1844 1631   1682 1407
Definitive
570 241 265   241 265
Possible
328 347 321   201 172
Definite CVD at LASA B
- - - - 308 -
Definite CVD at LASA B or C
- - - - - 373

 

Appendix 1: Part of the GP questionnaire of 1992/93 (D.kriegsman and B. Penninx)
(PDF - In Dutch).