Article Text

Download PDFPDF

Rose angina predicts 23-year coronary heart disease mortality in women and men aged 40–49 years
  1. S Graff-Iversen1,
  2. R Selmer1,
  3. M-L Løchen2,3
  1. 1
    Norwegian Institute of Public Health, Nydalen, Oslo, Norway
  2. 2
    Institute of Community Medicine, University of Tromsø, Tromsø, Norway
  3. 3
    Department of Cardiology, University Hospital of North Norway, Tromsø, Norway
  1. Dr S Graff-Iversen, Norwegian Institute of Public Health, Marcus Thranesgate 6, Box 4404 Nydalen, NO-0403 Oslo; sgri{at}


Objective: To determine the long-term coronary heart disease (CHD) mortality in women and men with symptoms, according to the Rose Angina Questionnaire at a relatively young age.

Design: Cohort study with the baseline survey conducted during 1974–8. Information on symptoms was collected by a short, three-item version of the Rose Angina Questionnaire. Participants were re-invited to a similar survey five years later and followed for mortality throughout 2000.

Setting: Three counties in Norway (the Norwegian Counties Study).

Participants: 16 616 men and 16 265 women aged 40–49 years and denying CHD in 1974–8.

Main outcome measure: CHD mortality during 23 years.

Results: By the end of follow-up 1316 men (7.9%) and 310 women (1.9%) had died from CHD, including 16% (66/406) of men and 4% (24/563) of women with Rose angina in 1974–8. Rose angina implied an elevated mortality from CHD with adjusted hazard ratios 1.50 (95% CI 1.16 to 1.93) in men and 1.98 (95% CI 1.30 to 3.02) in women. According to calculations based on the Cox model these increases in risk are similar to those associated with elevations of total cholesterol by 1.8 mmol/l (men) and 2.5 mmol/l (women) or elevations of systolic blood pressure by 21 mm Hg (men) or 31 mm Hg (women).

Conclusions: Angina symptoms in ages as low as 40–49 years were associated with elevated long-term CHD mortality in Norwegian women and men. This indicates that the three-item version of the Rose Angina Questionnaire, although a screening tool rather than a diagnostic test, adds information on undiagnosed CHD in both sexes.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

The World Health Organization angina questionnaire (Rose Angina Questionnaire) has been used since the 1960s as a method for assessing the burden of angina pectoris in populations.1 2 Prevalence of angina symptoms and subsequent coronary heart disease (CHD) mortality by use of a shortened, three-item version of this screening instrument were studied in Norway in the 1960s and 1970s, at the height of the post-second world war CHD epidemic.36 Among men, studies have consistently found a strong positive association of positive outcome (Rose angina) with CHD mortality.5 79

Although the prevalence of Rose angina in women is rather high,10 11 the reliability of this symptom is low with the greatest discrepancies at younger ages.12 Studies have found a low predictive value of Rose angina with respect to short-term CHD mortality in women,5 13 but the long-term cardiovascular morbidity and mortality were found to be elevated in 45–64-year-old women by use of the original WHO Rose Angina Questionnaire in west Scotland.14 The aim of this study was to examine the long-term CHD mortality in a population cohort of nearly 33 000 Norwegian men and women who were in their 40s and free of recognised CHD when answering a short version of the Rose Angina Questionnaire.


The Norwegian Counties Study

During 1974–8 all men and women aged 35–49 years, living in three mostly rural counties in different parts of Norway, were invited to take part in a cardiovascular health survey.10 15 The response rate was 91% in total and 95% when discounting subjects who gave notice of inability to participate.10 The participants are considered to be representative of the non-urban population of Norway; however, they were not nationally representative, as none of the larger cities were included in this survey. For this study we selected the age group 40–49 years, born during 1925–36. The participants were re-invited to a similar survey in 1977–83, approximately five years after the baseline, using the same version of the Rose Angina Questionnaire.

At both surveys the participants filled in a questionnaire, reporting on myocardial infarction (MI), angina, stroke, diabetes, use of antihypertensive medication, smoking, some dietary habits, physical activity and some other factors relevant for health. Women were interviewed about menopausal status. A short version of the WHO Rose Angina Questionnaire was included by using the following formulation:

  1. Do you get pain or discomfort in your chest when walking up hills, stairs or hurrying on level ground? (Yes or no)

  2. If you get pain or discomfort in the chest when walking, do you usually stop? Slow down? Carry on at the same pace? (Mark the alternative best fitting)

  3. If you stop or slow down, does the pain disappear after less than 10 minutes? Or after 10 minutes or more? (Mark the alternative best fitting)

Rose angina was defined by the answer “yes” to the first question, “stop” or “slow down” to the second and “less than 10 minutes” to the third. The original version included the localisation of chest pain or discomfort to sternum or to the left chest and arm as an additional criterion.1 4 5 14 A Norwegian study has shown that the six-year mortality figures according to the short version and the original were practically identical, although the prevalence of Rose angina was twice as high when omitting the localisation criterion.5

At the health examination height and weight were measured to the nearest centimetre or 0.5 kg. Body mass index (BMI) was calculated as weight in kg divided by height in metres squared. Systolic and diastolic blood pressures (SBP and DBP) were measured in the sitting position twice after a 4-minute rest using a mercury sphygmomanometer. The second measurement was used in this analysis. Non-fasting serum total cholesterol (TC) concentration was measured with a non-enzymatic method.15

In all, 17 012 men and 16 425 women in age 40–49 years participated in the 1974–8 survey. In analyses of mortality, 37 men and 44 women with incomplete data and 359 men and 116 women with positive self-reports of MI or angina were excluded, giving 16 616 men and 16 265 women free of known CHD for inclusion. Of these, 15 318 men and 15 301 women (93%) took part in the survey in 1977–83. Data on education and income according to a full-scale population census in 198016 were obtained from Statistics Norway and linked to the survey file.

Mortality follow-up

Information about vital status and emigration throughout 2000 and causes of death according to the Norwegian Causes of Death Register were linked to the survey file by a personal identification number. The Causes of Death Register is based on medical death certificates and supplementary information, which are routinely notified in Norway, and coded at national level according to the International Statistical Classification of Diseases (ICD).17 Until 1986 the Norwegian edition of ICD-8 was used, in 1986–95 the Norwegian edition of ICD-9 was applied, and from 1996 the English edition of ICD-10 was implemented.

We used the underlying (main) cause of death. Cardiovascular disease (CVD) was defined as ICD-8 codes 390-444.1 or 444.3-459, ICD-9: 390-459 and ICD-10: I00-I99. For defining CHD we used ICD-8: 410-414, ICD-9: 410-414 and ICD-10: I20-I25. Sudden death of unknown origin was defined by ICD-8: 783 or 795, ICD-9: 798 or R96 in ICD-10. Mortality from chronic lung disease was defined by ICD-8: 491-493 or 518, ICD-9: 490-494 or 496 and J40-J47 in ICD-10 and from cancer by ICD-8: 140-209, ICD-9: 140-208 and ICD-10: C00-C97. No attempt was made to check the death certificates against hospital records.

Statistical analysis

Risk factors in 1974–8 were compared by use of the t test and χ2 test. Years of observation were calculated for all subjects from the date of examination to the date of death or emigration or until 31 December 2000 for those still alive and living in Norway. With the exception of 46 men and 53 women who emigrated all participants were registered alive or dead in the population register at the end follow-up.

We estimated mortality hazard ratios (HR) with 95% confidence intervals (CI) in men and women separately comparing people fulfilling the three Rose angina criteria with all who did not by use of the Cox proportional hazards model. Years of observation were used as the time scale. The assumption of proportional hazards was checked by plotting log minus log functions. Potentially confounding factors were included as covariates. Current daily smoking, known diabetes, overweight (BMI ⩾25 kg/m2), family history of CHD, lung disease symptoms (morning cough or phlegm), use of antihypertensives, no regular physical exercise and menopausal status were included with “yes” or “no”, while age, TC and SBP were included as continuous covariates. Educational attainment was dichotomised into elementary level (9 years) or higher (10 years plus). Low income was defined in men by personal income below median and in women by family income below median. Participants with missing data on education (n = 727, 2.1% of Rose angina negative and 5.1% of Rose angina positive) or income (n = 537, 1.5% of Rose angina negative and 4.5% of positive) were classified to the lowest level.

We first analysed the mortality from 1974–8 throughout 2000 by Rose angina in 1974–8, including Rose angina and age (model 1); these variables plus daily smoking (model 2) and other potentially confounding factors as additional covariates (model 3). In order to quantify the strength of the association of Rose angina with mortality compared to other risk factors, we used the Cox model to calculate the difference in SBP and TC which conferred the same HRs as Rose angina (positive versus negative).

Secondly, observation years were calculated from the 1977–83 survey throughout 2000 for those taking part in both surveys. In analysis of mortality from the second survey and onwards, Rose angina in 1974–8 together with self-reported CHD (MI or angina) and potential confounders registered in 1977–83 were included, with the aim of examining the impact of Rose angina in 1974–8 when given the updated covariates. Rose angina in 1974–8 was included together with age (model 1), with age and CHD status by 1977–83 (model 2) and with additional potentially confounding factors by 1977–83 (model 3). Analyses were performed using the statistical package SPSS 14.0 for Windows (SPSS Inc, Chicago, IL, USA).

Ethical considerations

Practically all participants gave written consent for their results to be sent to their doctor. With this background, permission to use their data in this project was granted by the Directorate of Social Affairs and Health by 13 May 2002. Concession for establishing a register of the cardiovascular health surveys was given by the Data Inspectorate by 7 April 1989 and 13 October 2003. Concession to handle the personal information from all registers in this project, according to relevant laws and regulations, was given by 10 May 2002. The authors had full access to the data and take responsibility for their integrity.


Among all men (n = 17 012) and women (n = 16 425) who participated in 1974–8, 3.5% (95% CI 3.2 to 3.8) of men and 3.9% (95% CI 3.6 to 4.2) of women, fulfilled the Rose angina criteria. Of 250 men and 106 women who reported angina pectoris, 60% of the men and 72% of the women screened positive for Rose angina. After exclusion of people with self-reports of MI or angina the prevalence of Rose angina was 2.4% (406/16 616; 95% CI 2.2 to 2.7) in men and 3.5% (563/16 265; 95% CI 3.2 to 3.7) in women, increasing by age from 1.5% at age 40 to 2.8% at age 49 in men and from 2.8% to 4.0% in women. Of those included as Rose angina negative 4% had discomfort by exertion, but did not meet the next two Rose angina criteria. Men and women with Rose angina differed from those without this symptom by several health indicators, however not by income (table 1).

Table 1 Cardiovascular risk factors by Rose angina at the 1974–8 survey

In men, 4046 (24.4%) died during follow-up; of these 1316 died from CHD. Among women, 2075 (12.8%) died, of these 310 died from CHD. Table 2 shows the crude mortality over 23 years. Among the men and women with Rose angina 16.3% (66/406) and 4.3% (24/563), respectively, died from CHD during follow-up. The CHD mortality rates were, in both sexes, 2.4 times higher with Rose angina compared to participants without this symptom. In men the mortality from non-coronary CVD was elevated by Rose angina by a factor of 2.88 (95% CI 2.00 to 4.14), while no significant difference was seen in women. Chronic lung disease showed elevated mortality by Rose angina, while cancer mortality did not (table 2).

Table 2 Mortality during 23 years by Rose angina in 1974–8: Number of deaths, deaths per 100 000 person-years and rate ratios (95% CI)

The median age at death from CHD in men with Rose angina was 59 years (range 41–72 years), and the first CHD death occurred in 1975. Among women the median age was 64 years (range 45–74 years) with the first CHD death in 1981.

Table 3 shows the associations of Rose angina with 23-year mortality from CHD. When including age, established CHD risk factors, lung symptoms and education the HRs were 1.50; 95% CI 1.16 to 1.93 in men and 1.98; 95% CI 1.30 to 3.02 in women. Calculation based on model 3 showed that Rose angina in men (HR 1.50) is associated with elevation of risk similar to 1.8 mmol/l increase in TC or 21 mm Hg increase in SBP, and in women the HR of 1.98 corresponded to elevations by 2.5 mmol/l and 31 mm Hg, respectively.

Table 3 Coronary heart disease mortality during 1974–2000 by Rose angina in 1974–8. Mortality hazard ratios (HR) from Cox regression models

We compared the 209 men and 240 women who developed chest symptoms on exertion with those who denied any chest discomfort on exertion. The age-adjusted HR for CHD mortality in men and women was 1.39; 95% CI 0.91 to 2.12 and 1.79; 95% CI 0.89 to 3.16, respectively, compared to 2.27; 95% CI 1.77 to 2.91 and 2.26; 95% CI 1.49 to 3.43 for those fulfilling all three Rose angina criteria (see table 3, model 1).

Mortality during 1977–2000

All included men and women were free of diagnosed CHD according to self-reports in 1974–8, but of the men with Rose angina in 1974–8, 13.4% (47/351) reported CHD (MI or angina pectoris) by the 1977–83 survey, compared with 2.0% (295/14 967) in men without this symptom. Among women these figures were 7% (37/530) and 0.8% (119/14 771), respectively. For those participating in both surveys we analysed CHD mortality in 1977–2000 by Rose angina in 1974–8, adjusting for updated information on CHD and risk factors by 1977–83 (table 4). In men, Rose angina in 1974–8 did not imply any excess risk when this updated information was taken into account, but in women the risk was elevated (HR 1.99; 95% CI 1.28 to 3.01).

Table 4 Mortality from coranary heart disease (CHD) during 1977–2000 by Rose angina in 1974–8

Cardiovascular and all-cause mortality

As shown in table 2, the crude CVD mortality rates during 1974–2000 were elevated in men and women with Rose angina with rate ratios of 2.50 (95% CI 2.04 to 3.07) in men and 1.89 (95% CI 1.36 to 2.64) in women. In men, age-adjusted HRs were 2.40; 95% CI 1.96 to 2.95 and 1.58; 95% CI 1.28 to 1.95 when adjusting for CVD risk factors (not shown in table). In women these HRs were 1.80; 95% CI 1.29 to 2.51 and 1.56; 95% CI 1.11 to 2.18, respectively. In men, non-coronary CVD mortality was increased by Rose angina with age-adjusted HR 2.75; 95% CI 1.91 to 3.95 and adjusted HR 1.91; 95% CI 1.31 to 2.77 when all covariates were included. No significant increase was seen in women (age-adjusted HR 1.30; 95% CI 0.75 to 2.27). However, women with positive screening for Rose angina in 1974–8 were more likely than those with negative screening to report of stroke in the 1977–83 survey (9/530 versus 40/14 771, p<0.001), while there was no significant difference in men.

The crude all-cause mortality was elevated by Rose angina in both sexes (table 2), and Cox models showed age-adjusted HRs 1.83; 95% CI 1.57 to 2.14 in men and 1.28; 95% CI 1.04 to 1.57 in women. When including smoking and other potentially confounding factors the HR was still elevated by 30% for men while no significant elevation was present for women.


In this cohort Rose angina in women and men aged 40–49 constituted a major risk factor for premature CHD death. For health monitoring purposes this indicates that a short version of the Rose Angina Questionnaire adds information on undiagnosed CHD in both sexes. In our cohort the increased risk by Rose angina is similar to the risk associated with a 1.8 mmol/l elevation of TC or 21 mm Hg elevation of SBP in men and to even higher elevations of these established risk factors in women.

Men were twice as likely as women to report a CHD five years after the first survey. Of women with Rose angina, 7%, only reported a CHD after five years, and angina symptoms in 1974–8 were predictive for later CHD mortality even when updated information on CHD and risk factors in 1977–83 was taken into account. One explanation could be that CHD in women during the first years of observation was less likely to have reached the stage of being clinically detectable, and another is that women were less likely than men to be thoroughly examined, given similar chest symptoms. However, a hypothesis of underused medical care in women could not be tested with our data.

Our study complies with the Renfrew-Paisley Study with respect to elevated long-term CHD mortality in both sexes.14 But the studies differ in several respects: we used a short version of the Rose Angina Questionnaire while the Scottish study used the original; our population was larger and younger at baseline; we did not include electrocardiograms; and we could not study hospitalisation during follow-up. In the Scottish cohort the age was 45–64 years at baseline and the prevalence of Rose angina was far higher than in our cohort. The prevalence difference would most probably have been larger if we had used the original screening instrument.

Women with Rose angina had an increased risk for non-coronary CVD events in the Scottish study.14 In our cohort this was the case for men only, although women with Rose angina were more likely than screening negatives to report stroke after five years. The all-cause mortality was elevated by Rose angina in both sexes in the Renfrew-Paisley Study, while we found a significant elevation only in men, when adjusting for risk factors. The younger age in our cohort, making a true mortality difference less detectable, should be considered. Together, the two cohort studies suggest that Rose angina is a risk factor for CVD and all-cause mortality in addition to being a risk factor for CHD in particular. Our data support the Renfrew-Paisley Study by indicating that Rose angina is not a benign finding and warrants investigation in both sexes.

In his considerations of sensitivity and specificity of the angina instrument, Geoffrey Rose noted that chronic bronchitis could be a reason for positive outcome.1 Later it has been confirmed that Rose angina may represent bronchial asthma.18 Rose angina was linked with lung symptoms and increased mortality from pulmonary disease in our cohort, although the absolute mortality from this cause was modest.

The strengths of our data include high participation rates in two population surveys repeated at five-year interval, high numbers of participants in a narrow age group and a practically complete follow-up. We used a short form of Rose Angina Questionnaire which has been validated against the original by mortality follow-up,5 and a study from Britain found that a shorter form than ours performed even better than the original version.19 The limitations of our study include the lack of clinical CHD examination at baseline and of morbidity data as end point. The survey questionnaire did not include depression, anxiety and non-cardiac pain symptoms which were associated with Rose angina in another Norwegian cohort.11 High density lipoprotein (HDL) and low density lipoprotein (LDL) cholesterol were not measured, but a large European project, of which our cohort is a part, concluded that a model based on TC gives similar risk estimates as a model based on TC/HDL cholesterol ratio.20 The validity of the causes of death can be questioned as around 10% of cases of death in Norway were validated by autopsy in the 1990s, but the percentage was higher in the case of CHD and in younger age groups.21

The Rose Angina Questionnaire has been used as a CVD health indicator for decades and is still in use.22 23 Cardiovascular consequences of Rose angina symptoms may differ in populations depending on CHD and CVD prevalence. The Rose Angina Questionnaire was developed as a screening tool when CHD incidence and mortality was, in Western countries, far higher than today in middle-aged people. But the lifetime CHD risk in most populations is still rather high, and we consider our results as relevant today.


In a long-term follow-up, angina symptoms in age 40–49 years, according to a short version of the Rose Angina Questionnaire, were associated with elevated CHD mortality in women and men. The relative excess risk conferred by such symptoms corresponded to the increase of TC by around 2 mmol/l or the elevation of SBP by 20–30 mm Hg. This indicates that angina symptoms add information on undiagnosed CHD in both genders.


We thank the staff at the former National Health Screening Service, Oslo, by now part of the Norwegian Institute of Public Health, who conducted the health surveys. We also thank Statistics Norway and its staff for our use of the census data and the Causes of Death Register.


View Abstract


  • Funding: No external support.

  • Competing interests: None.

  • Contributors: SGI initiated the analysis, analysed data and drafted the manuscript. RS gave methodological advises. RS and MLL contributed to result interpretation and critical review.