Breathlessness limiting exertion in very old adults: findings from the Newcastle 85+ study

Abstract

Introduction

Long-term breathlessness is more common with age. However, in the oldest old (>85 years), little is known about the prevalence, or impact of breathlessness. We estimated breathlessness limiting exertion prevalence and explored (i) associated characteristics; and (ii) whether breathlessness limiting exertion explains clinical and social/functional outcomes.

Methods

Health and socio-demographic characteristics were extracted from the Newcastle 85+ Study cohort. Phase 1 (baseline) and follow-up data (18 months, Phase 2; 36 months, Phase 3; 60 months, Phase 4 after baseline) were examined using descriptive statistics and cross-sectional regression models.

Results

Eight hundred seventeen participants provided baseline breathlessness data (38.2% men; mean 84.5 years; SD 0.4). The proportions with any limitation of exertion, or severe limitation by breathlessness were 23% (95% confidence intervals (CIs) 20–25%) and 9% (95%CIs 7–11%) at baseline; 20% (16–25%) and 5% (3–8%) at Phase 4. Having more co-morbidities (odds ratio (OR) 1.34, 1.18–1.54; P < 0.001), or self-reported respiratory (OR 1.88, 1.25–2.82; P = 0.003) or cardiovascular disease (OR 2.38, 1.58–3.58; P < 0.001) were associated with breathlessness limiting exertion. Breathlessness severely limiting exertion was associated with poorer self-rated health (OR 0.50, 029–0.86; P = 0.012), depression (beta-coefficient 0.11, P = 0.001), increased primary care contacts (beta-co-efficient 0.13, P = 0.001) and number of nights in hospital (OR 1.81; 1.02–3.20; P = 0.042).

Conclusions

Breathlessness limiting exertion appears to become less prevalent over time due to death or withdrawal of participants with cardio-respiratory illness. Breathlessness severely limiting exertion had a wide range of service utilisation and wellbeing impacts.

Key Points

• In the oldest old (>85 years), little is known about the prevalence or impact of long-term breathlessness.

• We found breathlessness limiting exertion is common in the oldest old but appears to become less prevalent over time due to death or deterioration of participants with cardio-respiratory illness. We also found it was associated with poorer self-rated health, depression, more primary care contacts and number of nights in hospital.

• Holistic breathlessness interventions are effective regarding symptoms, quality of life and health service utilisation but clinical trials include few oldest old. These interventions should be considered for the oldest old, but further research conducted to evaluate whether and how interventions should be tailored for this group.

Introduction

Chronic persistent breathlessness [1] is associated with reduced quality of life [2, 3], social [4] and workplace activity [5], increased health service utilisation [6] and a poorer prognosis [7], including reduced 2- and 10-year survival in those over 70 [8]. For older adults, multi-morbidity is the norm [9]; chronic breathlessness may be an added burden to living with multiple long-term conditions with effects on mobility, activities of daily living, social connectedness and independence in their own homes.

Long-term breathlessness is frequently neglected despite available interventions [10]. Published prevalence estimates in the general population vary between 9 and 59%, depending on the definition used and population studied; the symptom is more prevalent in older people and in women [7, 11–14]. Prevalent conditions such as lung disease, heart failure and cancer [15] become more common with age and cause chronic and acute-on-chronic [16] breathlessness.

Although the prevalence of chronic breathlessness increases with age amongst older adults, we reported a negative association with the oldest old having a lower prevalence than the youngest old [17]. The Newcastle 85+ Study [9], which aimed to explore the health and service use of a cohort of the oldest old, allows exploration of its impact in this often neglected group.

We investigated the extent to which, (i) medical condition(s) and other characteristics explain the presence of breathlessness limiting exertion; and (ii) breathlessness limiting exertion explains clinical, social and functional outcomes including healthcare utilisation.

Methods

The Newcastle 85+ study cohort

In this secondary data analysis, data relating to breathlessness and health and socio-demographic characteristics were extracted from the population-based Newcastle 85+ Study cohort, the methods of which have been reported elsewhere [9]. Eligible participants were consenting very old adults born in 1921, aged 85 in 2006, living in Northeast England registered with participating general practices in Newcastle or North Tyneside and providing written informed consent [9]. For the first multi-dimensional health assessment, data were collected between June 2006 and October 2007. The subsequent phases of the data collection took place 18 months (Phase 2), 36 months (Phase 3) and 60 months (Phase 4) after baseline (Phase 1). More details about retention from Phase 1 to Phase 4 are presented in the flowchart (Online Supplemental Figure 1).

The Newcastle 85+ Study was approved by the Newcastle and North Tyneside Local Research Committee One (Ref: 06/Q0905/2).

Hypotheses

In people aged 85 and over:

1. Breathlessness limiting exertion becomes less prevalent as the oldest old grow older.

2. Variables such as higher age, higher physical activity levels and more social participation are inversely related to breathlessness limiting exertion. Other variables, such as number and type (e.g. heart and lung diseases) of medical conditions, smoking/ex smoking and inflammation are positively related to breathlessness limiting exertion.

3. Chronic breathlessness is associated with poorer clinical, social and functional outcomes.

Dataset for analysis

Phases 1, 2, 3 and 4 were included in the descriptive analysis of ongoing participation by disease (Figure 1). Phases 1, 3 and 4 were included in the analysis of prevalence of breathlessness at each. Phase 2 data were not included because questions on self-reported breathlessness were administered to <50% of participants who provided baseline breathlessness data.

Figure 1.

Survival percentages by phase, disease status and severe breathlessness (n = 810–817 at baseline). Participants who withdrew their consent at any stage count as missing from that point on.

Phase 1 and Phase 3 formed the dataset for the multivariable analyses. Data from Phase 4 were not included due to insufficient sample size.

Phase 3 cross-sectional analyses were not feasible in some models, due to a lower sample size due to death (n = 227; 27.8%) or withdrawal (n = 105; 12.9%) compared to those who provided breathlessness data at Phase 1.

Analysis

Outcome variables: Hypotheses 1 and 2

Binary breathlessness variables

We derived two binary variables relating to two levels of breathlessness over the past 4 weeks; any breathlessness limiting exertion and breathlessness severely limiting exertion (‘any breathlessness’ and ‘severe breathlessness’, see Box 1).

Box 1.

Survey questions for limiting chronic breathlessness and categorisations in our study with study hypotheses.

Survey questions relating to breathlessness outcome variable: 1. So in the last 4 weeks, has shortness of breath limited your ability to move around your home (on one level)? [Possible answers: Yes/No/Limited for reason(s) unrelated to shortness of breath]2. (if so) How much has shortness of breath limited your ability to move around your home (on one level)? [Possible answers: A bit/A lot/Completely unable to move around the home due to shortness of breath]3. In the last 4 weeks, has shortness of breath limited your ability to walk outdoors on the level, at your own pace? [Possible answers: Yes/No/Limited for reason(s) unrelated to shortness of breath]4. (if so) How much has shortness of breath limited your ability to walk outdoors, on the level, at your own pace? [Possible answers: A bit/A lot/Completely unable to walk outdoors, on the level, at own pace due to shortness of breath]Categorisations in our study:  Binary variable (hypotheses 1 and 2) 1. Any breathlessness limiting exertion (‘any breathlessness’): (Yes to 1 AND/OR Yes to 3) = breathless; rest = not breathless2. Breathlessness severely limiting exertion (‘severe breathlessness’): ([Yes to 1 AND (‘A lot’ or ‘Completely unable. . .’) to 2] AND/OR [Yes to 3 AND (‘A lot’ or ‘Completely unable. . .’) to 4]) = breathless; rest = not breathless

To address hypothesis 1, the prevalence of the two levels was estimated for Phases 1, 3 and 4 and presented within 95% confidence intervals (CIs). Using data regarding deaths and withdrawals, the proportions of those with and without severe breathlessness and those with and without cardiovascular, lung disease or cancer still participating in Phases 2, 3 and 4 were calculated.

Outcome variables: Hypothesis 3

Self-rated health

Self-rated health is an overall assessment of physical and psychosocial health and a good indicator of health status and subsequent morbidity and mortality [18, 19].

Depression

Depression was measured using the Geriatric Depression Scale (GDS) total score [20]. Depression is prevalent in older adults and associated with limiting breathlessness in those over 70 [19]. Values <5 indicate no depression, whereas values over 10 almost always indicate depression.

Health service utilisation

Health service utilisation was measured by the number of primary care team attendances, and the number of nights spent in hospital over the previous 12 months. Chronic breathlessness in the general adult population is associated with increased attendance in primary and secondary care [6].

Explanatory variables: Hypothesis 2 and 3

Bivariate analyses were conducted to explore the relationship between explanatory and outcome variables for the relevant hypotheses. The candidate variables primarily encompassed sociodemographic, health status and health service utilisation (for detail see Online Supplemental Tables 1, 3 and 4). Of particular note, the Timed Up and Go (TUG) test [21] was included as, although no direct association has been published between the TUG and breathlessness in older adults, population studies show reduced mobility in those with severe breathlessness [3]. In addition, an association between chronic breathlessness and changes in the hypothalamic–pituitary–adrenal axis has been observed [22, 23], therefore the biomarkers cortisol and HS CRP were included.

For Hypothesis 2, the breathlessness variable was included as a binary outcome variable (see Box 1). For Hypothesis 3, the breathlessness variable was included as a three-category explanatory variable: (i) Breathlessness severely limiting exertion (see above, ‘severe breathlessness’); (ii) Breathlessness limiting exertion, but not severely (see above, ‘mild–moderate breathlessness’); (iii) Those who answered ‘no’ or had diminished mobility due to reasons unrelated to breathlessness, were classified as not having limiting breathlessness (‘no breathlessness’).

Statistical methods

As many variables passed a threshold signifying a statistically significant relationship of P < 0.2, the number was reduced to avoid overfitting given the small sample size in Phase 3. Candidate variables were prioritised according to plausible explanations supported by the literature. Sex and age were included into each model, despite lack of association with breathlessness in the bivariate analyses.

Apart from descriptive statistics, the analyses concerning Hypotheses 2 and 3 were performed using multiple regression models. Concerning Hypothesis 2, logistic regression was used due to the binary nature of the outcome variables. Concerning Hypotheses 3, ordinal and linear regression models were applied due to the categorical and continuous nature of the outcome variables.

Results

Breathlessness data were provided by 817 participants at baseline (38.2% men; average age 84.5 years; SD 0.4 years). Most (78% in both Phases) self-reported good to excellent health. Phase 1 mean (3.6, SD 2.6) and median (3, IQR 2–5) values for the GDS were below the threshold for possible depression. The average number of self-reported illnesses was greater than 1 (Phase 1: mean 1.7, SD 1.4; Phase 3: mean 2.1, SD 1.5). TUG times indicated a population with limited mobility, at risk of falls and frailty (Phase 1: mean 19 s, SD 15 s, median 14 s, IQR 11 s to 20s; Phase 3: mean 22 s, SD 18.8 s, median 17 s, IQR 12–24 s). Consistent with TUG times, self-reported physical activity indicated that although most reported mildly energetic physical activity at least once a week (74%), this dropped to 35% for moderate and 5% for very energetic physical activity. Most (83%) were not current drivers at Phase 1, rising to 88% by Phase 3.

Participants had contacted primary care on average 10 times in the previous year (Phase 1), increasing to 11.4 times in the previous year in Phase 3. Mean cortisol levels were higher than mean values in middle- and advanced age community-dwelling adults [24], although within the range. Mean high specificity C-reactive protein (Hs-CRP) levels were moderately raised consistent with systemic inflammation [25].

For detailed descriptive statistics of all included variables, see Online Supplemental Tables 2–4.

Hypothesis 1. Prevalence

The proportion of participants with any breathlessness at baseline was 23% (95% CIs 20–25%). At Phase 4, it was 20% (16–25%) (Online Supplemental Table 2). The numerical decrease was more marked for those with severe breathlessness (9%; 95% CIs 7–11% at baseline: 5%; 3–8% at Phase 4) (Online Supplemental Table 1).

Of ongoing participants, the proportion with cancer, cardio-respiratory disease or severe breathlessness numerically decreased over time, e.g. of those with a history of cancer at baseline, 42% participated in Phase 4 compared with 52% of those without a history of cancer. This pattern was most marked for severe breathlessness (29.5% with the condition vs. 52.2% without it in Phase 4) (Figure 1).

Hypothesis 2. Variables associated with breathlessness limiting exertion

In the multivariable analysis (Table 1), having more illnesses (odds ratio (OR) 1.34, 95% CIs 1.18–1.54; P < 0.001), self-reported respiratory (1.88, 1.25–2.82; P = 0.003) or cardiovascular disease (2.38, 1.58–3.58; P < 0.001) were significantly associated with any breathlessness. Associated variables of severe breathlessness only were the same but with higher likelihood (more illnesses: 1.51, 95% CIs 1.26–1.82; P = 0.001; respiratory: 3.07, 1.73–5.45; P < 0.001; cardiovascular: 2.74, 1.39–5.40; P = 0.004) and cortisol levels (0.998; 0.996–0.995; P = 0.015). In Phase 3, the strongest association was seen for respiratory disease. For all models, the percentage of variation explained was at most 24%, indicating other factors at play.

Table 1.

Associations with breathlessness limiting and severely limiting exertion: Logistic regression models^a

	Model 1: Phase 1 (n = 712)				Model 2: Phase 1 (n = 712)				Model 3: Phase 3 (n = 378)
	Outcome: Any breathlessness				Outcome: Severe breathlessness				Outcome: Any breathlessness
	n	OR	95% CIs	P	n	OR	95% CIs	P	n	OR	95% CIs	P
Age	712	1.17	0.76, 1.79	0.472	378	1.42	0.74, 2.72	0.286	378	1.09	0.59, 2.03	0.777
Sex
Male Female	283	1			283	1			142	1
	429	1.24	0.83, 1.85	0.297	429	1.48	0.80, 2.75	0.210	236	1.10	0.61, 1.97	0.754
Smoking status  a
Never Ex-smokers Smokers	260	1		0.881	260	1		0.881
	405	1.41	0.93, 2.14	0.107	405	1.01	0.54, 1.89	0.966
	47	1.42	0.64, 3.13	0.389	47	1.32	0.43, 4.10	0.627
Higher education
No Yes	628	1			628	1			316	1
	84	0.85	0.46, 1.57	0.606	84	0.38	0.11, 1.33	0.129	62	0.98	0.46, 2.08	0.957
No. self-reported illnesses (excl. breathlessness)	712	1.35	1.18, 1.54	<0.001	712	1.51	1.26, 1.82	<0.001	378	1.27	1.06, 1.52	0.010
Cancer
No Yes	611	1			611	1			305
	101	0.76	0.44, 1.32	0.333	101	0.90	0.40, 1.99	0.790	73	0.85	0.43, 1.70	0.652
Respiratory disease
No Yes	534	1			534	1			274
	178	1.88	1.25, 2.82	0.003	178	3.07	1.73, 5.45	<0.01	104	2.49	1.43, 4.34	0.001
Cardiovascular disease
No Yes	317	1			317	1			153	1
	395	2.38	1.58, 3.58	<0.01	395	2.74	1.39, 5.40	0.004	225	2.15	1.19, 3.88	0.011
TUG (seconds)	712	1.00	0.99, 1.01	0.965	712	1.01	0.99, 1.02	0.311	378	1.01	0.99, 1.02	0.357
HS CRP	712	1.01	1.00, 1.02	0.156	712	1.01	1.00, 1.03	0.093	378	1.01	1.00, 1.03	0.133
Cortisol	712	1.00	1.00, 1.00	0.833	712	1.00	0.99, 1.00	0.015	378	1.00	0.99, 1.00	0.053

^aModels for breathlessness severely limiting exertion were not included due to an insufficient number of events per variable. R² (Nagelkerke): 0.148 for Model 1; 0.244 for Model 2; 0.157 for Model 3.

BL, breathlessness; CIs, confidence intervals; CV, cardiovascular disease; HS CRP, high specificity C-reactive protein; TUG, timed up and gotest; OR, odds ratio; n, number; p, p-value.

Hypothesis 3. The relationship between breathlessness limiting exertion and loneliness, self-rated health, depression and health service utilisation

Severe breathlessness was associated with poorer self-rated health (OR 0.50, 029–0.86; P = 0.012), but not loneliness. The association with mild–moderate breathlessness with self-rated health was not significant (OR 0.85, 0.57–1.27; P = 0.424; Table 2). Due to the high number of variables in the models presented in Table 1, the model with severe breathlessness as outcome had to be restricted only to the baseline, given the lower sample size and the particularly low prevalence rates of severe breathlessness in Phase 3.

Table 2.

Associations with loneliness and self-rated health—Ordinal regression models at phase 1. Higher values indicate more frequent loneliness and better health, respectively

	Model 1: Loneliness (n = 724)					Model 2: Self-rated health (n = 721)
	n	OR	95% CIs	P		n	OR	95% CIs	P
DEPENDENT VARIABLE:   Feeling lonely					DEPENDENT VARIABLE:   Self-rated health
Never	412				Poor or Fair	153
Sometimes	245				Good	269
Often or Always	67				Very good	215
					Excellent	84
INDEPENDENT VARIABLES:
Age	724	0.78	0.53, 1.13	0.181		721	0.95	0.69, 1.31	0.759
Number of self-reported illnesses^	724	1.05	0.93, 1.18	0.445		721	0.66	0.59, 0.74	<0.001
GDS score (depression)	724	1.42	1.32, 1.53	<0.001		721	0.84	0.79, 0.90	<0.001
TUG (seconds)	724	0.98	0.97, 0.99	0.010		721	0.98	0.97, 0.99	0.002
Breathlessness
No breathlessness	553	1				550	1
Mild–moderate breathlessness	105	0.70	0.44, 1.11	0.125		105	0.85	0.57, 1.27	0.424
Severe breathlessness	66	0.97	0.55, 1.71	0.922		66	0.50	0.29, 0.86	0.012
Sex
Male	285	1				282	1
Female	439	1.92	1.31, 2.82	0.001		439	0.94	0.67, 1.32	0.732
Smoking status
Never						261	1
Ex-smokers						412	1.20	0.88, 1.63	0.248
Smokers						48	0.78	0.43, 1.41	0.408
Higher education
No	638	1				635	1
Yes	86	0.76	0.46, 1.27	0.300		86	1.38	0.90, 2.13	0.143
Living alone
No	279	1				277	1
Yes	422	4.83	3.32, 7.04	0.000		421	1.34	0.99, 1.82	0.060
Not applicable (care home, etc.)	23	3.89	1.51, 10.01	0.005		23	3.33	1.41, 7.88	0.006
Driving
No	587	1				586	1
Yes	137	1.20	0.74, 1.93	0.462		135	1.00	0.67, 1.49	0.995
Very energetic physical activity
No	682	1				679	1
Yes	42	1.56	0.76, 3.22	0.225		42	1.35	0.72, 2.50	0.350
Moderately energetic physical activity
No	446	1				445	1
Yes	278	1.34	0.90, 1.97	0.146		276	1.37	0.98, 1.93	0.067
Mildly energetic physical activity
No	139	1				138	1
Yes	585	0.70	0.44, 1.09	0.117		583	1.44	0.95, 2.19	0.086

Note. Test of parallel lines: P = 0.090 for Model 1; P = 0.514 for Model 2. Pseudo R²: 0.300 for Model 1; 0.298 for Model 2. CIs, confidence intervals; OR, odds ratio; n, number; p, p-value.

The association with mild–moderate breathlessness with self-rated health was not significant (OR 0.85, 0.57–1.27; P = 0.424; Table 2). Mild–moderate breathlessness was associated with depression, and severe breathlessness was associated with more primary care contacts (Table 3) and number of hospital nights (Table 4). In the repeated cross-sectional analyses, any breathlessness was associated with depression. At Phase 3, breathlessness was not significantly associated with the number of primary care contacts or hospital nights although the point estimate for severe breathlessness showed higher ORs (Tables 3 and 4).

Table 3.

Associations with GDS score and contacts with primary care team members in the last 12 months—Linear regression models, Phases 1 and 3

	Dependent variable: GDS score
	Model 1: Phase 1 (n = 724)					Model 2: Phase 3 (n = 388)
	B	SE B	Β	P	R 2	B	SE B	β	P	R 2
Mild–moderate breathlessness	0.81	0.24	0.11	0.001		0.68	0.32	0.10	0.038
Severe breathlessness	0.80	0.30	0.09	0.009		0.71	0.48	0.07	0.144
Female sex	0.10	0.20	0.02	0.626		−0.14	0.27	−0.03	0.610
Higher education	0.17	0.26	0.02	0.524		−0.03	0.31	0.00	0.927
Driving	−0.15	0.24	−0.02	0.532		−0.99	0.34	−0.15	0.004
Very energetic physical activities	−0.38	0.37	−0.04	0.313		−0.20	0.74	−0.01	0.786
Moderately energetic physical activities	−1.22	0.20	−0.24	<0.001		−1.17	0.29	−0.21	<0.001
Mildly energetic physical activities	−0.72	0.24	−0.11	0.003		−1.12	0.30	−0.21	<0.001
Living alone	0.03	0.18	0.01	0.887		0.46	0.26	0.09	0.080
Living in a care home, etc.	−0.88	0.52	−0.06	0.089		0.12	0.64	0.01	0.852
Age	0.12	0.19	0.02	0.529		−0.21	0.26	−0.04	0.418
Number of self-reported illnesses (excl. breathlessness)	0.30	0.06	0.18	<0.001		0.08	0.08	0.05	0.323
TUG (seconds)	0.01	0.01	0.05	0.159		0.00	0.01	0.03	0.584
R 2  unadjusted					0.224					0.207
R 2  adjusted					0.210					0.179
	Dependent variable: Contacts with primary care team members in the last 12 months
	Model 3: Phase 1 (n = 725)					Model 4: Phase 3 (n = 387)
	B	SE B	Β	P	R 2	B	SE B	β	P	R 2
Mild–moderate breathlessness	−0.38	0.82	−0.02	0.641		0.08	1.17	0.00	0.943
Severe breathlessness	3.54	1.04	0.13	0.001		2.11	1.71	0.06	0.219
Female	−0.37	0.63	−0.02	0.564		−0.57	0.91	−0.03	0.535
Higher education	−0.90	0.88	−0.04	0.307		0.04	1.11	0.00	0.968
Living alone	−0.07	0.62	0.00	0.906		−0.89	0.92	−0.05	0.337
Living in a care home, etc.	−3.02	1.71	−0.07	0.077		0.09	2.23	0.00	0.968
Ex-smokers	−0.39	0.62	−0.03	0.527
Smokers	−2.99	1.19	−0.10	0.012
Age	0.06	0.65	0.00	0.926		−1.55	0.93	−0.08	0.096
Number of self-reported illnesses (excl. breathlessness)	0.69	0.21	0.13	0.001		1.09	0.29	0.19	0.000
TUG (seconds)	0.00	0.02	0.00	0.911		0.05	0.03	0.09	0.074
GDS score	0.20	0.12	0.06	0.105		0.22	0.17	0.07	0.209
R 2  unadjusted					0.071					0.078
R 2  adjusted					0.055					0.053

GDS, Geriatric Depression Scale; TUG, timed up and go; OR, odds ratio; CIs, confidence intervals; n, number; p, p-value.

Table 4.

Predictors of number of hospital nights in the last 12 months—Ordinal regression models, Phases 1 and 3

	Model 1: Phase 1 (n = 724)				Model 2: Phase 3 (n = 390)
	n	OR	95% CIs	P	n	OR	95% CIs	P
DEPENDENT VARIABLE:   Number of nights in hospital last 12 months.
0 nights	571				302
1–6 nights	68				34
7+ nights	85				54
INDEPENDENT VARIABLES:
Age	724	1.03	0.67, 1.57	0.898	390	1.50	0.86, 2.63	0.157
Number of self-reported illnesses (excl. breathlessness)	724	1.10	0.96, 1.25	0.158	390	1.09	0.92, 1.30	0.293
GDS score	724	1.12	1.04, 1.20	0.003	390	1.09	0.99, 1.21	0.069
TUG (seconds)	724	1.02	1.01, 1.03	0.003	390	1.02	1.00, 1.04	0.014
Breathlessness
Mild–moderate breathlessness	105	1.08	0.64, 1.80	0.781	57	0.85	0.42, 1.73	0.657
Severe breathlessness	65	1.81	1.02, 3.20	0.042	24	1.28	0.51, 3.16	0.600
No breathlessness	554	1.00			309	1.00
Sex
Male	285	1.00			148	1.00
Female	439	0.67	0.45, 1.00	0.051	242	0.89	0.51, 1.54	0.680
Higher education
No	638	1.00			328	1.00
Yes	86	1.12	0.63, 1.97	0.698	62	0.85	0.42, 1.72	0.647
Living alone
No	278	1.00			132	1.00
Yes	423	1.33	0.89, 2.00	0.168	243	1.57	0.88, 2.80	0.127
Not applicable (care home, etc.)	23	0.97	0.30, 3.17	0.959	15	1.52	0.42, 5.43	0.521
Smoking status
Never	263	1.00
Ex-smokers	412	1.31	0.87, 1.98	0.198
Smokers	49	0.75	0.32, 1.74	0.504

Notes. Test of parallel lines: P1 P = 0.330; P3 P = 0.091. Notes: Test of parallel lines: P = 0.330 for Model 1; P = 0.091 for Model 2. Pseudo R²: 0.087 for Model 1; 0.069 for Model 2.

GDS, Geriatric Depression Scale; TUG, timed up and go; OR, odds ratio; CIs, confidence intervals; n, number; p, p-value.

Discussion

This is the first study of the prevalence and impact of breathlessness limiting exertion in the oldest old, the fastest growing subgroup in the population by percentage change [26]. The cohort study underlying the data provides a valuable source of insights into this burgeoning population [27, 28]. Data demonstrate that the proportion of people experiencing breathlessness reduces over time. Most likely, this is due to mortality or deterioration in people with more severe breathlessness. In this setting, breathlessness is a harbinger of death given the aetiologies that underlie it, consistent with other large population studies that span a wider age range [17, 42]. Those without long-term breathlessness are more likely to outlive their contemporaries with breathlessness. Breathlessness limiting exertion is associated with having more long-term illnesses. Any breathlessness limiting exertion was also associated with depression and, if exertion was severely limited by breathlessness, it was associated with poorer self-rated health, more primary care visits and more nights in hospital. Levels of physical activity were inversely associated with depression in a dose-related pattern. Self-rated health was inversely associated with breathlessness, depression and the TUG.

Previous work amongst older adults has been in younger groups and shows a higher prevalence of breathlessness [13]. Our data suggest that the findings of decreasing prevalence very late in life are likely due to deterioration, withdrawal or death of those with breathlessness-causing illnesses [17]. Age-related physical changes affecting lung capacity may also contribute to breathlessness in the very old (reduced peri-airway supportive tissue [29], chest wall compliance [30] and diaphragmatic strength [31]), but the net effect appears to be disease related.

Variables associated with breathlessness

Consistent with other studies, having more illnesses, and specifically lung and heart disease [15], was associated with breathlessness limiting exertion. We found no association with sex. Sex differences in one study of breathlessness disappear when adjusted for absolute lung volumes [32]. If lung volume sex differences become smaller in the very old [33], our sample size may have been insufficient to detect breathlessness differences. Likewise, although point estimates showed increased odds of breathlessness with smoking and education, this was not significant, reflecting a smaller contribution in the oldest old. Previous work demonstrated that once adjusted for other social determinants, education is not associated with health outcomes in older adults [34, 35].

Impact of breathlessness

The ‘dose-dependent’ association between breathlessness and primary and secondary health service utilisation is documented in a general adult population [6]. Breathlessness severely limiting exertion is associated with both increased primary care contacts and more nights in hospital. The increased use of GPs has been previously noted in this age-group, but the relationship with breathlessness was not explored [36].

The relationship between breathlessness in the general adult population and mobility has been described [3]. However, although the central importance of maintaining mobility for the physical and mental well-being of older adults is established [37, 38], the link with breathlessness as a possible important contributing factor has had little or no attention. There appears to be a complex interplay between physical exertion and ensuing breathlessness, which leads to a vicious cycle of avoiding activity to avoid breathlessness. This deconditioning results in accelerated muscle loss, further worsening breathlessness triggered by less and less exertion [39].

The wider consequences of depression [2], anxiety [2], social withdrawal [4] and loss of role form another vicious cycle which can aggravate breathlessness through emotional triggers. These connected cycles are well-described in the Breathing–Thinking–Functioning clinical framework for holistic breathlessness management [39]. Holistic breathlessness management, including psychosocial and physical exercise interventions targeting these vicious cycles, reduces hospital nights and depression in clinical trials [40].

Strengths and limitations

This was a secondary analysis that was conceived after the data were collected. These data provide missing information about a complex interplay between disease, ageing and breathlessness in the oldest old. Given the pseudo R² values, other factors are at play that will need to be explored. Although in the multivariate regression models all the dependent variables were mutually adjusted, we did not calculate interactions between cardiovascular and respiratory diseases on the prevalence over time.

Implications for clinical practice and research

Breathlessness services improve breathlessness, psychosocial wellbeing and help to facilitate more judicious use of health service utilisation by addressing the vicious downward cycle of reduced mobility, deconditioning, social interaction and mental health [40]. To date, clinical trial populations and health services research rarely focuses on the oldest old. Community-based long-term illness care should include routine enquiries about breathlessness-related limitations, currently often invisible due to a lack of inquiry [10], with appropriate breathlessness management. Equally, when breathlessness is identified, clinicians should inquire about other potential long-term health problems. Current knowledge about limiting breathlessness, its impact and benefits from interventions is based on data from younger populations; further study in the oldest old would help develop tailored interventions for this group.

Conclusions

Breathlessness limiting exertion affects between one in four and five oldest adults, becoming less prevalent over time. Breathlessness severely limiting exertion was associated with more primary care contacts and hospital nights, depression and worse self-reported health. Holistic breathlessness interventions may improve service utilisation and wellbeing; further study in the oldest old would help develop such interventions tailored for this group.

Declaration of Conflicts of Interest

None declared.

Declaration of Sources of Funding

The Newcastle 85+ Study has been funded by the Medical Research Council, Biotechnology and Biological Sciences Research Council, the Dunhill Medical Trust and the National Institute for Health Research School for Primary Care. Parts of the work have also been funded by the British Heart Foundation, Unilever Corporate Research, Newcastle University and NHS North of Tyne (Newcastle Primary Care Trust).

Lukas Pitel and Cynthia Forbes were supported by a grant from Yorkshire Cancer Research: University of Hull endowment release ‘Reducing Inequalities in Cancer Outcomes in Yorkshire: Realising our potential for innovation in Patient Management, Survivorship and Palliative Care Research.’ HEND405.

No additional funds were provided for this analysis.

Data Availability

Data may not be shared with anyone who is not listed in the data sharing agreement with the Newcastle 85+ study researchers. Requests to share the data with other people must be in writing and will be considered by the Data Guardians Group.