Lifestyle factors and experience of respiratory alarm symptoms in the general population

Study design and population

This study was a nationwide cohort study of 100 000 adults aged 20 years or older, randomly selected from the general population. The sample was drawn from the Danish Civil Registration System (CRS), in which all Danish citizens are registered with a unique personal identification number. The CRS contains information about every Danish resident's date of birth, gender, migration, etc.24 Each individual received a postal letter explaining the purpose of the study. A unique 12-digit login for a secure webpage was included in the letter, which provided access to a comprehensive web-based questionnaire. Participants without access to a computer, tablet or smartphone, were offered the opportunity to complete the survey as a telephone interview. A reminder letter was sent to the non-respondents 2 weeks after the invitational letter, and after an additional 2 weeks, the non-respondents were contacted by telephone and encouraged to participate.25

The questionnaire

A comprehensive questionnaire concerning the experience of 44 predefined specific and non-specific cancer alarm symptoms, as well as general and frequent symptoms, was developed. The questionnaire was based on standard rating scales, previously validated questionnaires and ad hoc items. The methodological framework for developing, pilot testing and field testing the questionnaire is described in detail elsewhere.25 Alarm symptoms for several different cancers (lung, gastrointestinal, urological and gynaecological) were selected based on a literature review that included national and international cancer referral guidelines.3 ,5 ,10 ,11 In total, four RAS indicative of lung cancer form the basis of this paper: prolonged coughing, shortness of breath, coughing up blood and prolonged hoarseness.11

Respondents were asked whether they had experienced one or more RAS within the preceding 4 weeks. The wording of the question regarding symptoms was: ‘Have you experienced any of the following bodily sensations, symptoms, or discomforts within the past 4 weeks?’. An item concerning when the symptom(s) occurred for the first time was also included. The response categories were: ‘Less than 1 month ago’, ‘1–3 months ago’, ‘3–6 months ago’ or ‘more than 6 months ago’. Questions regarding current smoking status, average alcohol intake, weight and height were also asked.

Statistical analyses

In the cancer referral guidelines, coughing and hoarseness are defined as RAS when they are prolonged, that is, last for more than 4–6 and 3–4 weeks, respectively.9 To approach concordance with these definitions, only respondents who had experienced the symptom for the first time more than 1 month ago were considered to have experienced prolonged coughing and prolonged hoarseness.

Covariates considered in the statistical analyses were gender, age, smoking status, alcohol intake and BMI. The respondents were divided into the following age groups: 20–39, 40–59, 60–79 and ≥80 years. Smoking status was categorised as current, former and never smoker. Alcohol intake was categorised in units per week: 0, 1–7, 8–21 and ≥22 units/week. BMI was calculated from height and weight, and categorised according to WHO guidelines as follows: underweight (BMI <18.5), normal weight (BMI 18.5–24.9), overweight (BMI 25–29.9) and obese (BMI ≥30).20

Prevalence estimates were calculated for reporting at least one RAS and for each individual RAS. One symptom was defined as having experienced at least one, but possibly more than one, of the four RAS. CIs for all prevalence estimates were calculated using the binomial distribution. Differences between reporting of RAS were tested for each covariate using either the χ² test or Fisher's exact test, as appropriate.

Logistic regression models were used to test for interaction between gender and each covariate with regard to at least one symptom, prolonged coughing, shortness of breath or prolonged hoarseness, respectively. The tests were made with Bonferroni adjustment to account for multiple testing. A test for interaction was not made for coughing up blood, because of the small number of respondents that reported this symptom.

To evaluate collinearity between lifestyle factors, correlation coefficients were calculated with Spearman's rank correlation. Multivariate logistic regression models were used to analyse the association between each covariate and reporting of at least one RAS, and each individual RAS. Owing to interactions, the analyses were stratified with respect to gender. Adjustments were made for possible confounders: age, smoking status, alcohol intake and BMI.

Linear trends in the reporting of RAS were tested with logistic regression models for age and alcohol intake. For alcohol intake, the test for trend only included respondents who drank more than 0 units/week.

All statistical tests were subjected to a significance threshold of 0.05. Data analyses were conducted using Stata IC 13 software.

Age

Odds of reporting at least one RAS, prolonged coughing, shortness of breath or prolonged hoarseness increased with increasing age (P_trend<0.001) among men (table 2) and among women (table 3).

Smoking status

Regarding both genders, odds of reporting at least one RAS were significantly higher for former smokers (OR_men=1.42, 95% CI 1.30 to 1.56; OR_women=1.25, 95% CI 1.15 to 1.36) and current smokers (OR_men=2.58, 95% CI 2.35 to 2.83; OR_women=2.45, 95% CI 2.25 to 2.68) than for never smokers. Current smoking was significantly associated with reporting prolonged coughing, shortness of breath and prolonged hoarseness for both genders (tables 2 and 3), and, among men, with coughing up blood (table 2).

Alcohol intake

Overall, individuals reporting an alcohol intake of 1–7 or 8–21 units/week had significantly lower odds of reporting RAS than those who reported drinking 0 units/week. Among both men and women who reported alcohol intake >0 units/week, we observed a trend of increased reporting of at least one RAS with increasing alcohol intake (P_trend<0.001). This was also the case for reporting prolonged coughing in both men and women (P_trend<0.001), and, among women, for prolonged hoarseness (P_trend=0.032; tables 2 and 3).

Body mass index

In men and women, odds of reporting at least one RAS were significantly higher for individuals who were underweight (OR_men=2.01, 95% CI 1.31 to 3.07; OR_women=1.33, 95% CI 1.07 to 1.64) or obese (OR_men=1.60, 95% CI 1.44 to 1.77; OR_women=1.66, 95% CI 1.51 to 1.84) compared to individuals at normal weight (tables 2 and 3). Being underweight or obese was also significantly associated with reporting prolonged coughing among both men and women, and, among women, for shortness of breath and prolonged hoarseness (tables 2 and 3).

Summary and main findings

This study shows that RAS are common in the general population. Sixteen per cent of respondents reported at least one symptom. The most prevalent symptoms were prolonged coughing (8.5%) and shortness of breath (8%), while prolonged hoarseness (3.4%) and coughing up blood (0.1%) were least frequent. Men more frequently reported RAS, and odds of reporting RAS increased with age for both men and women. Lifestyle factors were significantly associated with experiencing RAS. Former and current smokers were more likely to report RAS than never smokers. Being underweight or obese was positively associated with reporting RAS. Among individuals who reported alcohol intake, odds of reporting RAS increased with increasing alcohol intake; however, individuals reporting an alcohol intake were less likely to report RAS than individuals with no alcohol intake.

Strength and limitations

A major strength of this study is the large study sample of 100 000 randomly selected Danish individuals. The response rate (52.2%) is similar to or exceeds that of previous population-based studies.1 ,18 ,26 Although more of the respondents were women and the respondents were slightly older than the non-respondents, the respondents were fairly representative of the general adult Danish population. However, we were unable to eliminate the possibility that the respondents could differ from the non-respondents regarding other parameters, which might include a risk of overestimating or underestimating the prevalence.

Willingness to respond to the questionnaire might depend on the presence of symptoms.27 ,28 If individuals with many RAS were more willing to answer, the prevalence estimates might have been overestimated. However, individuals with many RAS might not have the surplus of energy to respond to a comprehensive questionnaire, which would counterbalance the aforementioned.

The fact that the questionnaire was web-based could have prevented some individuals, for example, the elderly, from participating in the survey.29 We sought to minimise this possibility by offering individuals without a computer, smartphone, or tablet, the opportunity to conduct the survey as a telephone interview. Nevertheless, a lower response rate was still found in the oldest age group, which might indicate that the telephone interview did not completely compensate for the possible selection. The lower response rate might result in bias, because older respondents might be in better health than older non-respondents. Thus, the prevalence of RAS among the oldest age group might be even higher than estimated in the present study.

Information about symptom experiences was self-reported, and respondents were asked to recall symptom experiences within the 4 weeks preceding completion of the questionnaire. The time of recall was chosen because it seems reasonable to assume that individuals can recall symptom experiences fairly accurately within that timespan.30 ,31 However, recall bias cannot be eliminated. Some respondents may misplace older symptoms in the specific time period, providing an overestimation of prevalence.32 However, others may recall fewer symptoms due to, for example, memory decay, providing an underestimation of the prevalence.33

Reporting of lifestyle factors might be biased due to a general tendency to underreport smoking status, alcohol intake and weight, and a tendency to over-report height.34–36 However, web-based questionnaires have been suggested to enhance the perception of privacy among respondents, increasing the reliability of answers regarding sensitive issues, such as lifestyle factors.37 ,38 Although avoidance of misclassification might not be possible, the questionnaire was comprehensive and considered a broad range of different symptom experiences and topics. This makes it unlikely that, for example, the experience of RAS addressed initially in the questionnaire has affected answers regarding lifestyle factors addressed later in the questionnaire. A possible misclassification would, therefore, be non-differentiated.39

The analyses were adjusted for age and each lifestyle factor.17 ,40 ,41 Comorbidity was considered a potential confounder, but was not included in the model because we found that it was more likely to be a mediator of the association between lifestyle factors and experience of RAS.

Discussion of results and comparison with the existing literature

Few studies have estimated the prevalence of respiratory symptoms in the general population.2 ,18 ,26 ,42 McAteer et al18 found the prevalence of coughing to be 17.8%, compared with 8.5% in the present study, while Svendsen et al2 observed a prevalence of coughing of 6.5%. One explanation for these differences might be the different time interval for reporting symptom experiences. McAteer et al18 did not restrict the duration of coughing, while the present study only included coughing that lasted longer than 4 weeks, and Svendsen et al2 included individuals who experienced coughing for more than 6 weeks. Prevalence estimates for shortness of breath and coughing up blood in the McAteer study were comparable to those in the present study.18 Petrie et al42 and Whitaker et al26 observed higher prevalence of symptom experiences. Petrie et al42 published a prevalence of coughing of 28.3% and a prevalence of shortness of breath of 13.2%; however, they collected data during flu and cold season. Whitaker et al26 reported a prevalence of persistent coughing of 20.3%, but used a broader definition of persistent coughing than the present study, and also included symptom experiences within the last 3 months.

In contrast to the findings of the present study, McAteer et al18 observed no gender difference in the prevalence estimates of coughing and shortness of breath. Furthermore, older age groups reported significantly lower odds of coughing and shortness of breath,18 while the present study demonstrated higher odds of reporting RAS in the oldest age groups. The age span differed between McAteer et al,18 who only included individuals aged 60 years or younger, and the present study, which included all individuals aged 20 years or older. However, this difference does not explain the different findings in the two studies.

The present study shows that RAS are common. RAS are defined as warning signs of lung cancer, but may be signs of more benign conditions, as well. Distinguishing between benign conditions and serious diseases such as cancer is difficult, and poses a challenge for physicians as well as the general population. Experiencing and interpreting symptoms is a complex process that involves several parameters.14 Qualitative studies have mentioned that individuals in the general population often consider age, former experienced symptoms and lifestyle factors in their interpretation of symptoms.16 ,43

In the present study, current and former smokers were more likely to report RAS than never smokers. The results were expected, considering that smoking is a risk factor for several respiratory diseases, thus likely leading to symptoms.17 ,41 Studies have shown that current smokers have a tendency to normalise their symptoms.44 ,45 If that is the case, then the odds of current smokers experiencing RAS might be underestimated in the present study.

Individuals drinking 1–7 and 8–21 units/week were less likely to report RAS than individuals who never drink alcohol. One possible explanation for this is that the never drinkers represent a group characterised by morbidity resulting in many symptoms and inability to drink alcohol. Another, possibly more plausible explanation, is that the never drinkers may represent a group that has made a deliberate choice of healthy living,46 and are thus more aware of symptom experiences and report more symptoms when asked during completion of a questionnaire. This theory is not supported in the current literature, but could be investigated in future research.

Underweight and obese individuals were more likely to report RAS than individuals with normal weight. One possible explanation, at least for shortness of breath, is that underweight and obese individuals experience more of a strain on their bodies than individuals at normal weight. Underweight or obese individuals might also be more aware of their bodies, resulting in remembering and reporting more symptom experiences when asked during completion of a questionnaire. These hypotheses have not been tested in previous studies, but could be addressed in future studies.

Conclusion and implications

This population-based study showed that 16% of the general population experienced at least one RAS within the preceding 4 weeks and that lifestyle factors influenced the experience of RAS.

The first step in the diagnosis of lung cancer is for individuals with RAS to recognise their symptoms. Knowledge about the prevalence of RAS in the general population and in subgroups with different lifestyles might be useful in the understanding of the diagnostic pathway of lung cancer, and may help policymakers to develop targeted campaigns. Although many people experience RAS, few are diagnosed with lung cancer. Nevertheless, it is necessary to seek healthcare in order for lung cancer to be diagnosed. Smoking, older age and being underweight or obese are positively associated with experiencing RAS. Whether these factors also influence healthcare-seeking when RAS are experienced is unknown. Future research should investigate healthcare-seeking behaviour among subgroups with different lifestyles.