Open Access
Open Peer Review

This article has Open Peer Review reports available.

How does Open Peer Review work?

Symptom experience and subsequent mortality: results from the West of Scotland Twenty-07 study

  • Alison M Elliott1Email author,
  • Philip C Hannaford1,
  • Blair H Smith1,
  • Sally Wyke2 and
  • Kate Hunt3
BMC Health Services Research20066:158

DOI: 10.1186/1472-6963-6-158

Received: 08 August 2006

Accepted: 11 December 2006

Published: 11 December 2006

Abstract

Background

Associations between symptom experience and mortality have rarely been investigated. One study has suggested that the number of symptoms people experience may be an important predictor of mortality. This novel and potentially important finding may have important implications but needs to be tested in other cohorts.

Methods

858 people aged around 58 years were interviewed by nurses in 1990/1 as part of the West of Scotland Twenty-07 Study. They were asked about the presence of symptoms in the last month from a checklist of 33 symptoms. Measures of morbidity included symptom type (respiratory, musculoskeletal, gastrointestinal, mental health, neurological, systemic) and symptom summary measures looking at the number and impact of symptoms (total number; number participants tended to have; number participants did not tend to have; number which restricted usual activities; number which led to GP consultation). Hazard ratios for thirteen-year all-cause mortality were calculated for symptom types, symptom summary measures, and self-assessed health with and without adjustment.

Results

On unadjusted analysis, and after adjusting for gender, socio-economic status and smoking, mortality was elevated in individuals reporting respiratory, systemic and mental health symptoms. After additional adjustment for chronic conditions and self-assessed health, only the association between mental health symptoms and mortality remained significant. On unadjusted analysis, and after adjusting for gender, socio-economic status and smoking, mortality was elevated in individuals with many (≥ 6) symptoms in four of the symptom summary measures examined. These relationships were no longer significant after additional adjustment for chronic conditions and self-assessed health. A clear trend of increasing mortality as self-assessed health became poorer was observed. This pattern remained statistically significant after adjustment for gender, socio-economic status, smoking, chronic conditions and the total number of symptoms experienced.

Conclusion

Symptoms often thought of as minor may have important consequences later in life especially for those reporting mental health-related symptoms or those experiencing many symptoms. In this study however, self-assessed health appeared to be a better predictor of mortality than the type or number of symptoms experienced, even when the tendency to have and impact of the symptoms were taken into account.

Background

There has been growing interest in recent years in researching symptoms in the community [1, 2], particularly since symptoms are drivers of health care utilisation and have been shown to be the principal reason for clinic visits in about half of all outpatient encounters [2]. Previous population-based studies of a range of symptoms [310] have provided useful information on the prevalence of symptoms within different socio-demographic groups, examined symptom burden and looked at the impact of symptoms in terms of quality of life, reduced functioning and use of health services. While some of these studies have examined the relationship between symptoms and the presence of chronic disease or self-assessed health, none of them have investigated the relationship with mortality.

Only one previous study has investigated the relationship between symptom burden and mortality [11]. Sha et al. found that individuals with higher symptom scores were significantly more likely to die over the subsequent year than individuals with no or few symptoms, even after adjustment for age, sex, race, chronic conditions, self-assessed health, depression and anxiety. This suggests that the number of symptoms people experience may be an important predictor of mortality. This novel finding, if replicated, may have important implications. Further evidence is needed however to see whether these findings exist in other studies. In addition, since a simple count of the number of symptoms experienced by individuals is a crude measure with limitations, further work is needed to investigate whether this relationship is stronger for more detailed symptom measures.

In order to explore this we examined, in a UK cohort study, the associations between type and number of symptoms and mortality. We were particularly interested in investigating the relationship between the burden of symptoms with different impact and mortality. We hypothesised that summary measures of symptom experience, which take into account the number of symptoms experienced, any tendency to have the symptoms and any impact on functioning, would be better predictors of mortality than simple counts of symptoms. In addition, since self-assessed health has been found to be a strong predictor of morbidity and mortality in previous studies [12, 13], we also examined the relationship between self-assessed health and mortality to see if the relationship is mediated through symptoms.

Methods

Participants and procedures

The data were from the Regional sample of the West of Scotland Twenty-07 study, a longitudinal study of health amongst three cohorts, aged 15, 35 and 55 when first studied in 1987/88 (see elsewhere for full details) [1416]. In brief, in order to get a representative sample of people in each age group a two stage sampling strategy was used. The sampling frame was the 1986 Voluntary Population Survey, an enhanced electoral register with details of names, sex and ages of all household members maintained by Strathclyde Regional Council (SRC). For ethical reasons SRC required written consent from potential participants to forward their names to the MRC for possible inclusion in the study. The University of Glasgow Ethics Committee for non-clinical research approved the West of Scotland Twenty-07 study.

Here we use data on the 55 year old cohort as there are few deaths to date in the younger cohort. For the 55 year old cohort 2155 people were approached by SRC and 1196 (55%) gave written consent to have their name transferred. Of these, 1042 (87%) agreed to participate in the study when approached. The cohort was first followed up in 1991 when 858 (82%) of the original 1042 men and women were interviewed at around 58 years of age. This contact included more detail on symptoms (such as whether or not activities were limited by the symptom) than the baseline survey.

Measures

During a nurse-led interview conducted in participants' homes, individuals were asked whether they had experienced any of 33 listed symptoms (including respiratory/ENT, musculo-skeletal, gastrointestinal, mental health, neurological, systemic (general) and other symptoms: Table 1). These symptoms were presented to participants on cards listing up to six symptoms at a time. Participants were asked whether they had experienced each symptom in the last month and whether it was a symptom they tended to have (i.e. one that they experienced often). For each symptom experienced in the last month participants were also asked whether it had caused them to cut down on their usual activities, and whether they had consulted a general practitioner for this occurrence of that symptom. Self-reports of longstanding chronic conditions were elicited using an extended version of the question used in the UK General Household Survey [17, 18]. Each condition was coded to a condition group (International Classification of Diseases chapter heading) according to the British Royal College of General Practitioners coding scheme [19]. Self-assessed health in the previous year was measured using the question: "In the last year how would you rate your general health – excellent, good, fair or poor?" A wide range of personal and social information was also collected including gender; occupational social class (non-manual, manual); housing tenure (owned/mortgaged, rented from council/housing association, privately rented and other); car ownership (no, yes); Carstairs deprivation category (1–2 (least deprived), 3–5, 6–7 (most deprived))[20]; and smoking status (never, ex, current).
Table 1

List of 33 symptoms inquired about as part of the West of Scotland Twenty-07 study*

Respiratory/ENT symptoms

Colds or flu

Persistent cough

Palpitations or breathlessness

Sinus, catarrh or blocked nose

Sore throat

Wheezy chest

Musculoskeletal symptoms

Sore or stiff back

Stiff or painful joints

Muscle tightness or tension

Gastro-intestinal symptoms

Constipation

Stomach pain or cramps

Diarrhoea

Indigestion

Poor appetite

Sickness or nausea

Piles or haemorrhoids

Mental health symptoms

Difficulty sleeping

Nerves or tension

Difficulty concentrating

Worrying over every little thing

Feeling depressed

Neurological symptoms

Headaches

Trembling hands

Fainting or dizziness

Systemic (general) symptoms

Feeling generally run down

Always feeling tired

Hot and sweating a lot

Cold sweats

Other symptoms

Trouble with ears

Trouble with eyes

Skin rash or skin problems

Trouble with teeth or mouth

Kidney or bladder problems

*Symptoms were not presented to participants in this order or in these groupings

Death notifications

Participants in the West of Scotland Twenty-07 Study were 'flagged' at the National Health Service central registries in Scotland and England at the start of the study in 1987/88. At regular intervals the study is notified of the date and cause of any deaths occurring in the cohort.

Data analysis

The data were analysed using SPSS for Windows. Deaths occurring in the cohort up to the end of May 2004 were included in the analysis and used to calculate hazard ratios for thirteen-year all cause mortality. In order to identify which factors were unadjustedly associated with mortality, unadjusted hazard ratios and their 95% confidence intervals were calculated for: gender; occupational social class; housing tenure; car ownership; deprivation category; smoking status; presence of a musculoskeletal condition; presence of a respiratory condition; presence of a digestive condition; presence of a cardiovascular condition; presence of a mental health condition; the total number of chronic conditions suffered and self-assessed health.

In order to examine the association between type of symptom (respiratory/ENT, musculoskeletal, gastrointestinal, mental health, neurological, and systemic) and mortality, unadjusted hazard ratios and their 95% confidence intervals were calculated for each symptom type, using absence of such symptoms as the referent group in each case. Cox regression was then used to adjust these hazard ratios for potential confounding from gender, socio-economic status, smoking, chronic conditions, and self-assessed health. We did not adjust for all of our measures of socio-economic status, since these measures were highly correlated. We used housing tenure only as the measure of socio-economic status in our adjustments.

In order to examine the association between number of symptoms and mortality, symptoms were categorised into five symptom summary measures: total number of symptoms experienced in the last month; number of symptoms experienced in the last month which participants tended to have; number of symptoms experienced in the last month which participants did not tend to have; number of symptoms experienced in the last month which resulted in cutting down on usual activities; and number of symptoms experienced in the last month which resulted in consulting a general practitioner. Within each summary measure, symptoms were aggregated so that individuals were categorised as having at baseline (1991) 0, 1, 2, 3–5, or ≥ 6 such symptoms (this division was made to ensure we had adequate numbers of individuals in each category). The exception to this approach occurred for the symptom summary measure indicating the number of symptoms experienced in the last month which participants did not tend to have, where individuals were aggregated as having at baseline 0, 1, 2, ≥ 3 such symptoms because of the small numbers of people with higher symptom counts in this summary measure. Unadjusted hazard ratios and their 95% confidence intervals were calculated for different levels within each symptom summary measure, using absence of such symptoms as the referent group in each case. Cox regression was then used to adjust these hazard ratios for potential confounding.

In order to examine the association between self-assessed health and mortality, unadjusted hazard ratios were calculated for poor, fair and good health, with excellent health as the referent group. Hazard ratios were then adjusted for potential confounding from gender, socio-economic status, smoking, chronic conditions, and total number of symptoms in the last month.

Results

A total of 170 deaths (19.8 percent of the cohort) occurred in the cohort in the thirteen years between 1991 and 2004. Most deaths were caused by cardiovascular disease (n = 62) or cancer (n = 56). Women were significantly less likely to have died than men, and car owners were significantly less likely to have died than those not owning a car (Table 2). Individuals in manual occupations, those who rented their house from the council or a housing association, those in higher deprivation categories and smokers were all significantly more likely to have died than those in the referent group for each variable.
Table 2

Hazard ratios for thirteen-year all cause mortality by gender, socio-economic factors and smoking

Socio-demographic factors

n in sample

Died n (%)

Unadjusted Hazard Ratio (95% CI)

Gender

   

Male

399

97 (24.3)

1.00

Female

459

73 (15.9)

0.63 (0.47 to 0.86)

Social class

   

Non manual

361

51 (14.1)

1.00

Manual

497

119 (23.9)

1.80 (1.29 to 2.49)

Housing tenure

   

Owned/mortgaged

474

62 (13.1)

1.00

Rented from council/housing association

352

100 (28.4)

2.35 (1.71 to 3.23)

Private rented and other

31

7 (22.6)

1.90 (0.87 to 4.15)

Car ownership

   

No

319

88 (27.6)

1.00

Yes

538

81 (15.1)

0.51 (0.38 to 0.69)

Deprivation category

   

1+2

135

17 (12.6)

1.00

3–5

407

72 (17.7)

1.43 (0.84 to 2.42)

6+7

314

81 (25.8)

2.17 (1.29 to 3.66)

Smoking status

   

Never

285

29 (10.2)

1.00

Ex

239

42 (17.6)

1.80 (1.12 to 2.89)

Current

334

99 (29.6)

3.24 (2.14 to 4.90)

Individuals with cardiovascular and respiratory conditions were more likely to have died than people without these conditions (Table 3). The presence of musculoskeletal, digestive or mental health conditions did not appear to affect mortality. There was a strong association between number of reported chronic conditions and mortality. A clear trend was seen between self-assessed health and mortality, with mortality increasing as self-assessed health became poorer (Table 3).
Table 3

Hazard ratios for thirteen-year all cause mortality by chronic conditions and self-assessed health

Chronic conditions

n in sample

Died n (%)

Unadjusted Hazard Ratio (95% CI)

Presence of musculoskeletal condition

   

No

439

90 (20.5)

1.00

Yes

419

80 (19.1)

0.91 (0.67 to 1.23)

Presence of respiratory condition

   

No

705

128 (18.2)

1.00

Yes

153

42 (27.5)

1.59 (1.12 to 2.25)

Presence of digestive condition

   

No

683

129 (18.9)

1.00

Yes

175

41 (23.4)

1.32 (0.93 to 1.87)

Presence of cardiovascular condition

   

No

627

95 (15.2)

1.00

Yes

231

75 (32.5)

2.45 (1.81 to 3.33)

Presence of mental health condition

   

No

730

138 (18.9)

1.00

Yes

128

32 (25.0)

1.37 (0.93 to 2.01)

Number of chronic conditions

   

0

107

11 (10.3)

1.00

1

211

23 (10.9)

1.07 (0.52 to 2.20)

2

222

48 (21.6)

2.18 (1.13 to 4.20)

3

147

33 (22.4)

2.33 (1.18 to 4.61)

≥ 4

171

55 (32.2)

3.69 (1.93 to 7.05)

Self-assessed health

   

Excellent

118

8 (6.8)

1.00

Good

398

53 (13.3)

2.01 (0.95 to 4.22)

Fair

238

66 (27.7)

4.55 (2.18 to 9.47)

Poor

103

42 (40.8)

7.72 (3.62 to 16.44)

The most commonly reported symptoms in the last month were stiff or painful joints (38.8%), difficulty sleeping (29.7%), headaches (25.8%), back trouble (23.1%) and always feeling tired (21.9%). For further details on symptom prevalence see elsewhere [8]. Table 4 shows the hazard ratios for thirteen-year all cause mortality by type of symptom. On unadjusted analysis participants with respiratory/ENT, mental health or systemic symptoms had a significantly higher mortality risk than those not reporting such symptoms. These results remained statistically significant after adjustment for socio-demographic factors. However, after additional adjustment for the presence of a respiratory or cardiovascular condition and self-assessed health, only mental health symptoms showed a significant association with mortality (HR, 95%CI: 1.42, 1.02 to 1.97).
Table 4

Hazard ratios for thirteen-year all cause mortality by type of symptom

Type of symptom

n in sample*

Died n (%)

Unadjusted Hazard Ratio (95% CI)

Hazard ratio (95% CI) adjusted for socio-demographic factors1

Hazard ratio (95% CI) adjusted for socio-demographic factors and chronic conditions2

Hazard ratio (95% CI) adjusted for socio-demographic factors, chronic conditions and self-assessed health3

Respiratory/ENT symptoms

      

Does not have symptoms

424

66 (15.6)

1.00

1.00

1.00

1.00

Has symptoms

434

104 (24.0)

1.60 (1.17 to 2.17)

1.44 (1.06 to 1.97)

1.24 (0.89 to 1.71)

1.10 (0.79 to 1.54)

Musculoskeletal symptoms

      

Does not have symptoms

418

78 (18.7)

1.00

1.00

1.00

1.00

Has symptoms

440

92 (20.9)

1.12 (0.83 to 1.51)

1.14 (0.84 to 1.55)

1.15 (0.84 to 1.56)

0.99 (0.72 to 1.35)

Gastro-intestinal symptoms

      

Does not have symptoms

552

101 (18.3)

1.00

1.00

1.00

1.00

Has symptoms

306

69 (22.5)

1.29 (0.95 to 1.76)

1.31 (0.96 to 1.78)

1.30 (0.95 to 1.77)

1.13 (0.82 to 1.56)

Mental health symptoms

      

Does not have symptoms

461

70 (15.2)

1.00

1.00

1.00

1.00

Has symptoms

397

100 (25.2)

1.78 (1.31 to 2.41)

1.79 (1.31 to 2.45)

1.66 (1.21 to 2.28)

1.42 (1.02 to 1.97)

Neurological symptoms

      

Does not have symptoms

581

112 (19.3)

1.00

1.00

1.00

1.00

Has symptoms

277

58 (20.9)

1.11 (0.81 to 1.53)

1.08 (0.78 to 1.49)

0.99 (0.71 to 1.37)

0.81 (0.58 to 1.13)

Systemic symptoms

      

Does not have symptoms

553

92 (16.6)

1.00

1.00

1.00

1.00

Has symptoms

305

78 (25.6)

1.63 (1.21 to 2.21)

1.50 (1.10 to 2.05)

1.28 (0.93 to 1.76)

1.00 (0.71 to 1.41)

* Numbers in each of the five symptom categories add to 858 the total number in the cohort. Numbers given are relevant for unadjusted hazard ratios. Numbers in the sample for each of the adjusted analyses varies only slightly since there was little missing data for the factors we have adjusted for (as can be seen from tables 1 and 2).

1 Adjusted for gender, housing tenure and smoking status

2 Adjusted for gender, housing tenure, smoking status and the presence of a respiratory or cardiovascular condition

3 Adjusted for gender, housing tenure, smoking status, the presence of a respiratory or cardiovascular condition and the number of symptoms present in the last month

The total number of symptoms experienced in the last month reported by respondents at baseline ranged from 0 to 21, with a median of two symptoms (Figure 1). Table 5 shows the hazard ratios for thirteen-year all cause mortality by number of symptoms for each of the five symptom summary measures examined. Four of the symptom summary measures (total number of symptoms; number of symptoms which participants tended to have; number of symptoms which restricted usual activities; and number of symptoms which resulted in consulting a general practitioner) showed very similar results. As the number of symptoms reported rose, the chances of dying tended to increase, although there was no clear trend in any of the summary measures with increasing number of symptoms. Individuals who reported having six or more symptoms in each of these four summary measures were significantly more likely to have died than those without such symptoms. These results remained statistically significant after adjustment for gender, socio-economic status and smoking. After additional adjustment for the presence of respiratory and cardiovascular conditions the hazard ratios were smaller and the results were of only borderline significance. After further adjustment for self-assessed health no significant associations remained. There was no relationship between number of symptoms and mortality in the fifth symptom summary measure (number of symptoms experienced in the last month which participants did not tend to have).
https://static-content.springer.com/image/art%3A10.1186%2F1472-6963-6-158/MediaObjects/12913_2006_Article_335_Fig1_HTML.jpg
Figure 1

Distribution of total number of symptoms experienced in the last month based on self report from a list of 33 symptoms.

Table 5

Hazard ratios for thirteen-year all cause mortality by number of symptoms in five symptom categories

  

n in sample*

Died n (%)

Unadjusted Hazard Ratio (95% CI)

Hazard ratio (95% CI) adjusted for socio-demographic factors1

Hazard ratio (95% CI) adjusted for socio-demographic factors and chronic conditions2

Hazard ratio (95% CI) adjusted for socio-demographic factors, chronic conditions and self-assessed health3

Number of symptoms in the last month

0

87

9 (10.3)

1.00

1.00

1.00

1.00

 

1

116

16 (13.8)

1.32 (0.58 to 2.99)

1.40 (0.62 to 3.17)

1.21 (0.53 to 2.75)

1.24 (0.54 to 2.82)

 

2

128

27 (21.1)

2.13 (1.00 to 4.54)

2.06 (0.96 to 4.41)

1.94 (0.90 to 4.15)

1.99 (0.93 to 4.29)

 

3–5

309

60 (19.4)

1.99 (0.99 to 4.00)

1.99 (0.98 to 4.03)

1.74 (0.85 to 3.55)

1.40 (0.68 to 2.86)

 

≥ 6

218

58 (26.6)

2.76 (1.37 to 5.58)

2.53 (1.24 to 5.16)

2.03 (0.99 to 4.19)

1.38 (0.65 to 2.89)

Number of symptoms in the last month which participants tended to have

0

120

15 (12.5)

1.00

1.00

1.00

1.00

 

1

141

18 (12.8)

0.99 (0.50 to 1.97)

1.06 (0.53 to 2.10)

1.00 (0.50 to 1.99)

0.99 (0.50 to 1.97)

 

2

138

28 (20.3)

1.68 (0.90 to 3.15)

1.66 (0.88 to 3.13)

1.62 (0.86 to 3.05)

1.51 (0.80 to 2.86)

 

3–5

291

59 (20.3)

1.70 (0.96 to 2.99)

1.60 (0.90 to 2.80)

1.44 (0.81 to 2.56)

1.14 (0.64 to 2.06)

 

≥ 6

168

50 (29.8)

2.56 (1.44 to 4.57)

2.23 (1.24 to 4.02)

1.85 (1.01 to 3.37)

1.15 (0.61 to 2.18)

Number of symptoms in the last month which participants did not tend to have

0

588

123 (20.9)

1.00

1.00

1.00

1.00

 

1

156

29 (18.6)

0.85 (0.57 to 1.28)

0.90 (0.60 to 1.36)

0.93 (0.61 to 1.40)

0.98 (0.65 to 1.49)

 

2

59

11 (18.6)

0.89 (0.48 to 1.64)

1.18 (0.63 to 2.20)

1.20 (0.64 to 2.25)

1.16 (0.62 to 2.18)

 

≥ 3

55

7 (12.7)

0.61 (0.28 to 1.30)

0.71 (0.33 to 1.53)

0.67 (0.31 to 1.44)

0.72 (0.33 to 1.55)

Number of symptoms in the last month which resulted in cutting down on usual activities

0

526

84 (16.0)

1.00

1.00

1.00

1.00

 

1

130

30 (23.1)

1.50 (0.99 to 2.28)

1.45 (0.95 to 2.21)

1.40 (0.92 to 2.14)

1.19 (0.77 to 1.82)

 

2

66

18 (27.3)

1.74 (1.03 to 2.93)

1.54 (0.90 to 2.61)

1.41 (0.82 to 2.41)

1.05 (0.60 to 1.83)

 

3–5

97

22 (22.7)

1.54 (0.96 to 2.46)

1.35 (0.84 to 2.18)

1.17 (0.72 to 1.90)

0.85 (0.51 to 1.40)

 

≥ 6

39

16 (41.0)

3.21 (1.88 to 5.49)

2.42 (1.39 to 4.20)

1.72 (0.97 to 3.06)

1.21 (0.67 to 2.21)

Number of symptoms in the last month which resulted in consulting the GP

0

457

72 (15.8)

1.00

1.00

1.00

1.00

 

1

166

39 (23.5)

1.55 (1.05 to 2.29)

1.49 (1.00 to 2.21)

1.48 (0.99 to 2.20)

1.38 (0.93 to 2.06)

 

2

91

19 (20.9)

1.39 (0.84 to 2.31)

1.46 (0.88 to 2.43)

1.36 (0.82 to 2.27)

1.13 (0.67 to 1.91)

 

3–5

96

24 (25.0)

1.63 (1.02 to 2.61)

1.51 (0.94 to 2.43)

1.34 (0.83 to 2.16)

1.08 (0.66 to 1.75)

 

≥ 6

48

16 (33.3)

2.51 (1.46 to 4.31)

1.96 (1.13 to 3.41)

1.58 (0.91 to 2.77)

1.13 (0.63 to 2.01)

* Numbers in each of the five symptom categories add to 858 the total number in the cohort. Numbers given are relevant for unadjusted hazard ratios. Numbers in the sample for each of the adjusted analyses varies only slightly since there was little missing data for the factors we have adjusted for (as can be seen from tables 1 and 2).

1 Adjusted for gender, housing tenure and smoking status

2 Adjusted for gender, housing tenure, smoking status and the presence of a respiratory or cardiovascular condition

3 Adjusted for gender, housing tenure, smoking status, the presence of a respiratory or cardiovascular condition and self-assessed health

Self-assessed health was strongly associated with mortality. Respondents with self-assessed poor health had a mortality risk over seven times (HR, 95% CI: 7.72, 3.62 to 16.44) that of participants reporting excellent health status. This association remained significant even after adjustment for gender, socio-economic status, smoking, chronic conditions and the total number of symptoms in the last month (adjusted HR, 95%CI for poor vs excellent self-assessed health: 4.46, 1.95–10.21).

Discussion

Summary of main findings

On unadjusted analysis, and after adjusting for gender, socio-economic status and smoking, mortality was elevated in individuals reporting respiratory, systemic and mental health symptoms suggesting that some types of symptoms have a greater impact on mortality than others. After additional adjustment for a history of respiratory and cardiovascular conditions and self-assessed health, only the association between mental health symptoms and mortality remained statistically significant. This finding suggests that the presence of mental health symptoms has important implications for subsequent health. On unadjusted analysis, and after adjusting for gender, socio-economic status and smoking, mortality was elevated in individuals reporting many (≥ 6) symptoms for four of the five symptom summary measures examined, suggesting that the presence of many symptoms (of different frequency and severity) in late mid-life predicts subsequent mortality. The absence of any association among individuals reporting symptoms in the last month that they did not tend to have highlights the importance of considering the tendency to have and/or impact of symptoms rather than simple symptom counts. After additional adjustment for chronic conditions and self-assessed health in the last year none of the associations between symptom groups and mortality remained statistically significant.

Comparison with existing literature

Few studies have examined the relationship between type of symptoms and mortality in community populations, tending, when done, to focus on respiratory and depressive symptoms. A significant relationship between respiratory symptoms (such as breathlessness, wheeze and cough) and all-cause mortality has consistently been found in previous population studies, even after adjustment for confounding factors [2125]. Our findings are broadly in line with these previous studies, although our association with mortality was lost after adjustment for the presence of respiratory and cardiovascular conditions and self-assessed health. Less consistent results have been found in population studies assessing the relationship between mental health symptoms and mortality. Vogt et al. [26] reported no relationship. Everson-Rose et al. [27] found a significant association after adjusting for age, gender and race. However, after additional adjustments for socio-economic status, lifestyle behaviours, functional limitations and chronic health conditions the relationship was eliminated. Wulsin et al. [28] found that all-cause mortality was directly associated with depressive symptoms, with higher mortality risks as depressive symptom scores increased. Our findings support Wulsin et al. and highlight that mental health symptoms may be important independent predictors of subsequent mortality in late mid-life.

Only one previous study has investigated the reporting of symptom burden in relation to mortality [11]. As part of a multi-centre study in the USA looking at 12 physical symptoms in 3498 adults aged 60 and over, Sha et al. found that individuals with higher symptom scores were significantly more likely to have died over the subsequent year than individuals with no or few symptoms. These results remained significant even after adjustment for age, sex, race, chronic conditions, self-assessed health, depression and anxiety. Our findings are broadly in line with Sha et al's, showing a two- to three- fold increase in mortality with increasing number of total symptoms reported when examined unadjustedly or after adjustment for socio-economic factors. Unlike Sha et al, however, our associations were smaller and no longer statistically significant after adjustment for self-assessed health. We examined a different, smaller population and different symptoms over a longer time period. Sha et al examined a smaller range of (physical only) symptoms, used simpler numeric summaries and did not include any measure of their impact. On the other hand, some of the symptoms they examined might be considered more serious (e.g. chest pain) than those examined in our study and this may explain why Sha et al. found a stronger relationship between symptoms and mortality than we did. In addition, they studied an older cohort of individuals (mean age at baseline 69 compared to 58 years in our study). Self-reported symptoms can arise from a variety of causes, including the early stages of disease which has yet to be detected, self-limiting minor illnesses which have no important long-term consequences, or from a known long-standing condition. The underlying mix of causes is likely to differ substantially with age; with older age groups more likely to have a higher percentage of symptoms related to chronic, serious or undetected conditions.

In our study, self-assessed health was strongly associated with mortality. This finding is in line with previous research, which has reported a two to seven fold increase in mortality among those with poor compared with excellent self-assessed health, after adjusting for clinical and medical history measures [12, 13]. Our findings suggest that self-assessed health in the last year may be a better predictor of future mortality than the number of symptoms experienced, even when the tendency to have and impact of these symptoms are taken into account. When assessing their own health individuals are likely to take into account a range of factors such as medical health status, functional ability, physical fitness, experience of symptoms, psycho-social well-being, lifestyle and health behaviours [29, 30]. A measure of symptom burden may be a sub-component of self-assessed health and therefore less able to predict mortality.

Strengths and limitations of this study

Strengths of this study were its population-based design, the large number and range of symptoms covered and the detailed social information collected. The study was able to examine underlying morbidity in terms of self-reported symptoms, chronic conditions and self-assessed health. Although we had a relatively small number of participants, we had a long period of follow-up, enabling us to look at subsequent mortality over a 13-year period. This was a longer follow-up than in previous research [11] and resulted in a larger number of deaths available for analysis, thereby increasing statistical power (although there were insufficient data to permit cause-specific mortality analyses). An important limitation was the fact that although we asked about a large number of symptoms (over 30), most were largely minor, self-limiting in nature. Different relationships may have been found if we had included more serious symptoms (e.g. chest pain, weight loss etc). In addition, the analyses were based on self-reported data, which were not cross-checked with medical records. The experience of symptoms however, is by definition subjective and many of the symptoms (and some of the longstanding chronic conditions) will not have been presented to general practice, hampering attempts to validate the self-reported symptoms against medical records. Participants in the analyses were all aged around 58 years at the time of interview. As already highlighted it is likely that different age groups will report different numbers of symptoms and chronic conditions. It is also likely that their perceived self-assessed health will be different, as indicated by previous studies [3133]. Whether symptom experience is related to mortality in other age groups, and whether self-assessed health would remain a stronger predictor of mortality than symptoms needs to be tested in other studies. Finally, our measures of symptom experience and self-assessed health had different time reference points (last month and last year respectively). It is noteworthy that self-assessed health over the last year appeared to be a stronger predictor of mortality than symptoms experienced in the previous month.

Conclusion

This study has shown that symptoms often thought of as 'minor' or self-limiting may have important consequences later in life, especially for those reporting mental-health related symptoms or those experiencing many symptoms. Further work is needed to ascertain how best to help health-care professionals differentiate between symptoms which have serious outcomes and those which do not. Our observation that a single question about self-assessed health is a better predictor of mortality than various combinations of symptom experience points towards more research into understanding the utility of this measure.

Declarations

Acknowledgements

We would like to thank the nurse interviewers and other colleagues on the Twenty-07 Study. Thanks are due to Sally Macintyre and Geoff Der who provided comments on an earlier draft. The Twenty-07 Study is funded by the Medical Research Council of Great Britain. AME is supported by a Research Career Development Fellowship, funded by the Wellcome Trust. KH is employed by the Medical Research Council of Great Britain.

Authors’ Affiliations

(1)
Department of General Practice and Primary Care, University of Aberdeen, Foresterhill Health Centre
(2)
Department of Nursing and Midwifery, University of Stirling
(3)
MRC Social and Public Health Sciences Unit, Glasgow University

References

  1. Kroenke K, Laine C, (Eds): Investigating Symptoms: Frontiers in Primary Care Research. Ann Intern Med. 2001, 134: 801-930.
  2. Kroenke K, Harris L: Symptoms research: A fertile field. Ann Intern Med. 2001, 134: 801-802.View ArticlePubMedGoogle Scholar
  3. Verbrugge LM: Triggers of symptoms and health care. Soc Sci Med. 1985, 20: 855-876. 10.1016/0277-9536(85)90343-0.View ArticlePubMedGoogle Scholar
  4. Verbrugge LM, Ascione FJ: Exploring the iceberg. Common symptoms and how people care for them. Med Care. 1987, 25: 539-569. 10.1097/00005650-198706000-00008.View ArticlePubMedGoogle Scholar
  5. Antonovsky H, Maoz B, Pilpel D, Arad T: Personal and health factors associated with frequency of visits to the primary care clinic. Fam Pract. 1989, 6: 182-187.View ArticlePubMedGoogle Scholar
  6. Kroenke K, Price RK: Symptoms in the community. Prevalence, classification, and psychiatric comorbidity. Arch Intern Med. 1993, 153: 2474-2480. 10.1001/archinte.153.21.2474.View ArticlePubMedGoogle Scholar
  7. Stoller EP, Forster LE, Portugal S: Self-care responses to symptoms by older people. A health diary study of illness behaviour. Med Care. 1993, 31: 24-42. 10.1097/00005650-199301000-00002.View ArticlePubMedGoogle Scholar
  8. Wyke S, Hunt K, Ford G: Gender differences in consulting a general practitioner for common symptoms of minor illness. Soc Sci Med. 1998, 46: 901-906. 10.1016/S0277-9536(97)00217-7.View ArticlePubMedGoogle Scholar
  9. Krantz G, Ostergren PO: Common symptoms in middle aged women: their relation to employment status, psychosocial work conditions and social support in a Swedish setting. J Epidemiol Community Health. 2000, 54: 192-199. 10.1136/jech.54.3.192.View ArticlePubMedPubMed CentralGoogle Scholar
  10. Molarius A, Janson S: Self-rated health, chronic diseases, and symptoms among middle-aged and elderly men and women. J Clin Epidemiol. 2002, 55: 364-370. 10.1016/S0895-4356(01)00491-7.View ArticlePubMedGoogle Scholar
  11. Sha MC, Callahan MD, Counsell SR, Westmoreland GR, Stump TE, Kroenke K: Physical symptoms as a predictor of health care use and mortality among older adults. Am J Med. 2005, 118: 301-306. 10.1016/j.amjmed.2004.12.015.View ArticlePubMedGoogle Scholar
  12. Idler EL, Benyamini Y: Self-rated health and mortality: a review of twenty-seven community studies. J Health Soc Behav. 1997, 38: 21-27. 10.2307/2955359.View ArticlePubMedGoogle Scholar
  13. Benyamini Y, Idler EL: Community studies reporting association between self-rated health and mortality: additional studies, 1995 to 1998. Res Aging. 1999, 21: 392-401.View ArticleGoogle Scholar
  14. Macintyre S, Annandale E, Ecob R, et al: The West of Scotland Twenty-07 Study: health in the community. Readings for a New Public Health. Edited by: Martin CJ, McQueen DV. 1989, Edinburgh: Edinburgh University Press, 56-74.Google Scholar
  15. Ford G, Ecob R, Hunt K, Macintyre S, West P: Patterns of class inequality in health through the lifespan: class gradients at 15, 35 and 55 years in the West of Scotland. Soc Sci Med. 1994, 39: 1037-1050. 10.1016/0277-9536(94)90375-1.View ArticlePubMedGoogle Scholar
  16. West P, Ford G, Hunt K, Macintyre S, Ecob R: How sick is the West of Scotland? Age specific comparisons with national datasets on a range of health measures. Scott Med J. 1994, 39: 101-109.PubMedGoogle Scholar
  17. Macintyre S, Ford G, Hunt K: Do women 'over-report' morbidity? Men's and women's responses to structured prompting on a standard question about chronic illness. Soc Sci Med. 1999, 48: 91-100. 10.1016/S0277-9536(98)00292-5.View ArticleGoogle Scholar
  18. Wyke S, Hunt K, Walker J, Wilson P: Frequent attendance, socio-economic status and burden or health. An investigation in the West of Scotland. Eur J Gen Pract. 2003, 9: 48-55.View ArticlePubMedGoogle Scholar
  19. Royal College of General Practitioners: The classification and analysis of general practice data. 1986, London: Royal College of General PractitionersGoogle Scholar
  20. Carstairs V, Morris R: Deprivation and health in Scotland. 1990, Aberdeen: University PressGoogle Scholar
  21. Ebi-Kryston KL: Respiratory symptoms and pulmonary function as predictors of 10-year mortality from respiratory disease, cardiovascular disease, and all causes in the Whitehall study. J Clin Epidemiol. 1988, 41: 251-260. 10.1016/0895-4356(88)90129-1.View ArticlePubMedGoogle Scholar
  22. Carpenter L, Beral V, Strachan D, Ebi-Kryston KL, Inskip H: Respiratory symptoms as predictors of 27 year mortality in a representative sample of British adults. BMJ. 1989, 299: 357-361.View ArticlePubMedPubMed CentralGoogle Scholar
  23. Vollmer WM, McCamant LE, Johnson LR, Buist AS: Respiratory symptoms, lung function, and mortality in a screening center cohort. Am J Epidemiol. 1989, 129: 1157-1169.PubMedGoogle Scholar
  24. Knuiman MW, James AL, Divitini ML, Ryan G, Bartholomew HC: Lung function, respiratory symptoms and mortality: Results from the Busselton health study. Ann Epidemiol. 1999, 9: 297-306. 10.1016/S1047-2797(98)00066-0.View ArticlePubMedGoogle Scholar
  25. Frostad A, Soyseth V, Anderen A, Gulsvik A: Respiratory symptoms as predictors of all cause mortality in an urban community: a 30-year follow-up. J Intern Med. 2006, 259: 520-529. 10.1111/j.1365-2796.2006.01631.x.View ArticlePubMedGoogle Scholar
  26. Vogt T, Pope C, Mullhooly J, Hollis J: Mental health status as a predictor of morbidity and mortality: a 15-year follow-up of members of a health maintenance organization. Am J Public Health. 1994, 84: 227-231.View ArticlePubMedPubMed CentralGoogle Scholar
  27. Everson-Rose SA, House JS, Mero RP: Depressive symptoms and mortality risk in a national sample: confounding effects of health status. Psychosom Med. 2004, 66: 823-830. 10.1097/01.psy.0000145903.75432.1f.View ArticlePubMedGoogle Scholar
  28. Wulsin LR, Evans JC, Vasan RS, Murabito JM, Kelly-Hayes M, Benjamin EJ: Depressive symptoms, coronary heart disease, and overall mortality in the Framingham Heart Study. Psychosom Med. 2005, 67: 697-702. 10.1097/01.psy.0000181274.56785.28.View ArticlePubMedGoogle Scholar
  29. Bailis DS, Segall A, Mahon MJ, Chipperfield JG, Dunn EM: Perceived control in relation to socio-economic and behavioural resources for health. Soc Sci Med. 2001, 52: 1661-1676. 10.1016/S0277-9536(00)00280-X.View ArticlePubMedGoogle Scholar
  30. Garrity TF, Somes GW, Marx MB: Factors influencing self-assessment of health. Soc Sci Med. 1978, 12: 77-81.PubMedGoogle Scholar
  31. Krause NM, Jay GM: What do global self-rated health items measure?. Med Care. 1994, 32: 930-942. 10.1097/00005650-199409000-00004.View ArticlePubMedGoogle Scholar
  32. Jylha M, Leskinen E, Alanen E, Leskinen AL, Heikkinen E: Self-rated health and associated factors among men of different ages. J Gerontol. 1986, 41: 710-717.View ArticlePubMedGoogle Scholar
  33. Idler EL: Age differences in self-assessment of health: age changes, cohort differences or survivorship?. J Gerontol. 1993, 48: S289-300.View ArticlePubMedGoogle Scholar
  34. Pre-publication history

    1. The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1472-6963/6/158/prepub

Copyright

© Elliott et al; licensee BioMed Central Ltd. 2006

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Advertisement