- Email: [email protected]
Inequalities in influenza vaccine uptake among people aged over 74 years in Britain
Preventive Medicine 41 (2005) 545 – 553 www.elsevier.com/locate/ypmed
Inequalities in influenza vaccine uptake among people aged over 74 years in Britain Punam Mangtani, M.Sc., M.D.T, Elizabeth Breeze, Ph.D., Sari Kovats, M.Sc., Edmond S.W. Ng, M.Sc., Jennifer A. Roberts, Ph.D., Astrid Fletcher, Ph.D. London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK Available online 1 April 2005
Abstract Background. In the UK, payments to providers (General Practitioners) for vaccinating all people aged over 64 years old against influenza commenced in 2000. Little information exists on the relationship between uptake and need. We assessed factors influencing uptake and equity in uptake in over 74 year olds. Methods. We analysed cohort data from 5572 subjects in a community-based trial. Analyses took into account clustering and location of general practices in terms of Underprivileged Area (UPA) Score and area Standardised Mortality Ratio (SMR). Results. Vaccine uptake in practices in the most deprived tertile was 0.88 (95% CI 0.80–0.96) that of the least deprived and mid tertile, adjusted for confounding. Within each deprivation tertile, uptake in the mid and highest SMR tertile was 0.86 (95% CI 0.79, 0.94) and 0.87 (95% CI 0.81, 0.95) that of the lowest respectively. Uptake was 10% lower at the most in individuals with poorer quality housing. Higher uptake if married or with respiratory conditions and lower uptake if smoked had cognitive impairment or depression did not explain the socioeconomic differentials. Conclusions. Lower uptake in practices in deprived areas supports targeting of resources. At the individual level, those who are more isolated require support to access influenza vaccination. D 2005 Published by Elsevier Inc. Keywords: Influenza vaccine; Vaccination utilisation; Socioeconomic factors; Elderly
Introduction In 2000, the influenza vaccination programme was extended to all those over 64 in the UK [1]. General practices in the UK, as the source of all ambulatory care, are the main providers of influenza vaccination. Funds were made available to identify and approach all those eligible, monitor uptake, and support general practices, with insufficient staff. General Practitioners also received an item of service payment for vaccinations. There was some concern that influenza vaccine delivery would be inversely related to need as with other preventive services. Information, however, on the equity of influenza vaccination receipt and factors that could be addressed to improve uptake amongst elderly people is limited. Uptake T Corresponding author. Fax: +44 207 636 8739. E-mail address: [email protected] (P. Mangtani). 0091-7435/$ - see front matter D 2005 Published by Elsevier Inc. doi:10.1016/j.ypmed.2005.02.001
amongst people aged 64 years and older was 61% nationally in 2000 [2]. An initial analysis in a study population of over 74 year olds indicated that uptake increased from 51% in 1999 to 64% in 2000, but with no significant change in the fact that those with greatest socioeconomic disadvantage were less likely to be vaccinated than others [3]. This study sought to determine whether wider factors, such as pre-existing ill health, depression, cognitive impairment, and social support influenced vaccine uptake and whether these accounted for differences by socioeconomic position. The vaccination services of practices, such as active recruitment, were also examined.
Methods We used data on patients aged over 74 years in the UK Medical Research Council (MRC) Trial of Assess-
546
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
ment and Management of Older People in the Community. The general practice was the unit of randomisation. The design and methods are described elsewhere [4]. The trial was conducted in 106 practices recruited through the MRC General Practice Research Framework (GPRF) and selected to be representative of the joint tertiles of the UPA area deprivation scores [5] and Standard Mortality Ratios (SMRs) for electoral wards in Britain. The tertiles of SMR were nested within the tertiles of the UPA score. The trial evaluated a package of multi-dimensional assessment and management of older people. One of the main components was the method of assessment—Universal or Targeted. People eligible for the bover 74Q health assessment, excluding those in nursing homes or terminally ill, were invited to participate. The analysis of influenza vaccine uptake was confined to patients in practices in the Universal Arm who were invited as only these patients had full information collected on underlying medical conditions and socioeconomic factors based on a detailed health and social assessment by a study nurse. The assessments took place between 1995 and 2000 depending on when the practice was recruited. Patients were followed up to assess if they still belonged to the practice in 2000 by flagging for deaths and embarkations at the National Health Service Central Register (NHSCR) and by practices notifying changes of GP. Data on individual flu vaccination status were requested from practices. They were also asked about organisational factors likely to affect vaccination rates including patient reminders, extent to which registers of serious diseases were held, and level of computerisation, i.e. whether consultations and information on underlying medical conditions were recorded. Patients included had to be registered with the practice on 30th November 2000 when 95% of vaccinations had been given. The trial patients did not include those in nursing homes. Housing tenure was classified into owner-occupation, rented, and supported accommodation (sheltered and residential home). Supported accommodation was treated separately because it does not necessarily reflect the previous socioeconomic position of the occupier. Central heating in the home was combined with housing tenure as a marker of socioeconomic position. As type of area may determine ease of access to vaccination, the Carstairs’ deprivation index for the Enumeration District (ED) in which the patient resided was used as a socioeconomic marker of the area [6]. Population density from census data (persons per square kilometre within 5 km radius of the centroid of the ED) was also used, categorised into: urban (over 2500 persons/km2); 1000–2500 persons/ km2; 250–999 persons/km2; and rural (b250 persons/km2). Apart from gender and age, the other potential variables considered were: (i) Practice factors: list size, number of General Practioners (GPs), UPA and SMR tertiles of the ward in
which the practice was located, previous fund-holder status (where some payments for secondary care was directed by practices as opposed to centrally), computerisation level, and practice recruitment policy. (ii) Composite physical health indicators as reported by patients: respiratory history (at least one of shortness of breath, chronic cough or phlegm indicating chronic bronchitis, wheeze, ever-diagnosed by a doctor as having pneumonia or emphysema or asthma); cardiovascular history (at least one of heart attack or stroke as diagnosed by a doctor or a history of possible angina); general frailty (at least one of poor or fair self-reported health compared to others of the same age, perceiving oneself as not very or not at all active, being in the lowest quintile of BMI, or unable to do two or more out of eight activities of daily living). (iii) Mental health or ability. Depression was assessed on the Geriatric Depression Scale [7]. The Mini Mental State Examination [8] was used to suggest some cognitive deficit. (iv) Current smoking status. (v) Social support, such as marital status/living with a partner, contact with people outside the household, and presence of a confidante. All analyses took into account the clustered nature of the sample by using robust estimates derived from the bsvyQ suite of commands available in Stata 7 software [9]. Multilevel modelling was not used because the average number of EDs per practice was small, many EDs only contained one person, and the interest was in average effects. Modelling of the outcome relative to socioeconomic position and the potential for explanation was carried out using Poisson regression without a time factor to obtain risk ratios for being vaccinated or not. Unless otherwise specified, analyses included UPA score and SMR tertiles as stratifiers. All models were adjusted for gender and age. First, models were constructed for groups of explanatory factors to see which were predictors. The significant component factors in these groups were then added into models in a sequence felt to reflect the pathways of influence on vaccination (Fig. 1). In order to see whether a longer time lapse between measurement of these factors and autumn 2000 could lead to a dilution of effect, interactions between time lapse and the socioeconomic factors were tested. We examined for whether the effects of socioeconomic position differed by gender or underlying health.
Results Response rate Valid data for 2000 were obtained from 64% (34/53) practices in the Universal Arm of the trial. Participating
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
547
Fig. 1. Framework for models, indicating the stage at which explanatory factors were added.
practices compared to all eligible practices were more likely to be previous fund-holders, to have a large list size, or to be in the best UPA and SMR tertile (Table 1). The analyses of socioeconomic and other factors for uptake were carried out on 5572 people with complete information from the nurse assessment on all factors. The derivation of this subgroup for analysis is given in Fig. 2. Vaccine uptake amongst all those with information on vaccine status (including people who had not participated in the nurse assessment) (n = 9152) was 64% (95% CI 60– 67) (for sex specific crude uptake, see Table 2). People who participated in the nurse assessment (n = 6643), adjusted for age and gender, were 1.3 (95% CI 1.2, 1.4) times more likely to receive a vaccination than non-participants (n = 2509).
Vaccine uptake in relation to practice and individual characteristics At the crude level, people in practices with either seven or more GPs or less than 5000 list size were less likely to be vaccinated than those in other practices. Recruitment policy and level of computerisation made no substantial difference to uptake (Table 2). Vaccination uptake, adjusted for age in regression models, was highest amongst people in owner-occupation with central heating and amongst those in urban areas (unadjusted vaccination percentages are given in Table 3). The pattern of uptake by Carstairs’ quintile was irregular but with some indication of lower uptake in the most deprived areas. Preliminary tabulations by gender, adjusted for age (Table 4), suggested that the groups least likely to be
548
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
Table 1 Ratio of practices participating in 2000 to all practices in the Universal Arm by practice characteristics All
Fund-holding
34/53
Yes No Scottish
List size 11/16 20/32 3/5
b5000 7499 9999 10,000+
No. GPs 11/19 11/17 5/7 7/10
vaccinated were those: without respiratory problems, some cognitive deficit, a smoker, not married, living alone or with son/daughter, not having a confidante, and some indication of depression for women. Vaccine uptake in male smokers was lower compared to non smokers. Neither a history of cardiovascular disease (CVD) nor general frailty was significant predictors. Although the factors associated with vaccine uptake were not identical for males and females, separate models for each gender showed that the final models were very similar so results are shown here with men and women combined. All the individual predictors also had an association with at least one of the socioeconomic measures (not shown). Factors that were independently statistically associated with vaccination within each group shown in Fig. 1 (adjusted for gender and age and allowing for clustering and stratification by practice UPA and SMR tertile) were retained for modelling of socioeconomic differentials. These were: number of GPs and list size, respiratory vulnerability, depression, cognitive deficit, smoking status, and marital status. The general pattern for personal socioeconomic position (as measured
2 3 4–6 7+
UPA tertile 10/14 9/14 10/18 5/7
Best Middle Worst
SMR tertile 13/17 9/16 12/20
Best Middle Worst
11/16 12/17 11/20
by housing tenure and central heating) as described below was unchanged by adjustment for UPA and SMR.
Modelling socioeconomic position and vaccine uptake Table 5 shows the results for socioeconomic/geographic factors of the cumulative models. Examining each socioeconomic factor separately suggested that vaccine uptake was highest amongst people in owner-occupied homes or in supported housing and in the most urban area (model 1). While there was variation in uptake across Carstairs’ deprivation quartiles, there was no clear trend. Controlling for Carstairs’ score, level of urbanisation, health status, and social support did not alter the general pattern of uptake related to personal socioeconomic position (models 2–6). Although the overall P value is only of the order of 0.10, the risk ratio estimates for people in private households other than owner-occupiers with central heating are all below 1.0 and similar to each other suggesting that collectively they are at modest disadvantage.
Fig. 2. Analysis sample of vaccination in 2000 compared with fuller Trial sample.
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
549
Table 2 Percentages of men and women vaccinated in the winter 2000 season by practice factors Practice factors
Men
All List size b5000 5000 or more Recruitment policy Active Passive Number of GPs 2–6 7+ Previous fund-holding status Fund-holder Not a fund-holder Scottish practice Computerisation of consultations Mostly manual Mostly computerised All computerised and some underlying conditions All computerised and all underlying conditions a
Women a
n
% (95% CI)
n
% (95% CI)a
3293
69.8% (66.5–72.8)
5859
60.5% (56.9, 64.1)
1759 410
71.8 (67.9, 75.4) 82.7 (77.9, 86.6) P = 0.001
2983 568
63.3 (59.3, 67.1) 72.2 (65.5, 78.0) P = 0.020
1122 676
75.2 (70.6, 79.2) 71.0 (64.3, 76.9) P = 0.36
2455 1096
65.3 (60.2, 70.0) 63.5 (58.7, 68.0) P = 0.67
1741 428
75.1 (71.3, 78.6) 68.7 (64.8, 72.3) P = 0.014
2791 760
66.5 (62.9, 69.9) 58.2 (53.6, 62.6) P = 0.002
826 1172 171
73.1 (67.4, 78.1) 75.1 (70.0, 79.6) 69.0 (60.8, 76.2) P = 0.41
1321 1884 346
63.3 (58.5, 67.9) 67.0 (62.1, 71.5) 57.8 (53.3, 62,2) P = 0.014
70.1 77.2 66.3 72.8
382 1487 419 1129
62.0 65.6 56.3 66.6
224 900 252 716
(62.5, (71.4, (54.3, (66.7,
76.7) 82.2) 76.4) 78.1) P = 0.41
(55.5, (58.4, (49.2, (61.4,
68.1) 72.1) 63.2) 71.4) P = 0.20
P values for practice factors from model adjusted for age.
The risk ratio for being vaccinated if in owner occupation without heating compared to those with central heating was 0.90 (0.85, 0.95) for people assessed 3 or more years previously and 1.14 (0.98, 1.33) for those more recently assessed, suggesting that there was no dilution of socio-
differentials with time since assessment. There were also no interactions between either gender or frailty and socioeconomic status (data not presented). After adjustment for personal socioeconomic position and other significant practice risk factors (list size and
Table 3 Percentages of men and women vaccinated in the winter 2000 season by personal socioeconomic position and geographic factors Men
Women
n
% (95% CI)a
n
% (95% CI)a
1710 231
76.0 (72.2, 79.4) 69.7 (61.9, 76.5)
2427 380
68.2 (64.4, 71.7) 58.7 (51.9, 65.2)
401 75 89
70.6 (65.3, 75.4) 64.0 (53.4, 73.4) 77.5 (68.9, 84.3) 0.089
744 139 364
62.8 (58.5, 66.8) 61.9 (51.4, 71.3) 62.4 (57.5, 67.0) 0.049
Carstairs’ index quintiles b Least deprived Second quintile Third quintile Fourth quintile Most deprived P value
729 666 531 288 133
73.1 (67.8, 76.7 (71.3, 72.3 (67.3, 77.8 (70.4, 63.9 (54.9, 0.078
77.8) 81.4) 76.8) 83.8) 72.1)
1081 1094 840 492 269
67.1 (62.4, 65.9 (60.8, 60.7 (56.2, 66.5 (58.6, 59.5 (51.0, 0.125
71.4) 70.7) 65.1) 73.5) 67.4)
Urban indicator c pers pkmsq Rural 250–1000 1000–2500 Urban P value
739 611 602 395
74.4 (68.0, 71.2 (63.4, 70.9 (65.0, 82.3 (76.5, 0.077
79.9) 77.9) 76.2) 86.9)
1183 1001 976 616
64.7 (58.7, 62.0 (56.2, 60.7 (54.8, 75.5 (71.1, 0.005
70.2) 67.5) 66.2) 79.4)
Personal socioeconomic position Owner-occupier Central heating None Rent Central heating None Supported P value
a b c
P value for heterogeneity from models adjusted for age; confidence intervals adjusted for sample design. Quintiles derived from distribution of Carstairs’ scores for EDs throughout Britain. Based on population density within 5 km radius of centroid of ED.
550
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
Table 4 Percentages of men and women vaccinated in the winter 2000 season by health status, smoking, and support Men b
History of cvd None P value History of respiratory problemsc None P value Indicator frailtyd Not P value Depression scoree More than 5 4–5 Less than 4 P value Cognitive deficit (MMSE)f Some None P value Cigarette smoking status Current Ex Never P value Marital status Married/co-habiting Widowed/divorced/ separated Single P value Living status Alone With spouse only With spouse and other With son/daughter With other P value Carer for someone else at home Yes No P value Social contact outside household Daily 2–3 times a week Less than twice a week Rarely P value Has confidanteg Spouse (and other) Other relative Other No confidante P value
Women % (95% CI)a
n
n
% (95% CI)a
579 77.6 (72.8, 81.7) 696 66.2 (60.7, 71,4) 1938 73.4 (69.6, 76.8) 3381 65.3 (61.9, 68.5) 0.082 0.451 1368 76.3 (71.9, 80.2) 2061 67.4 (63.9, 70.8) 1088 71.5 (68.9, 74.2) 1853 63.5 (59.3, 67.50) 0.014 0.006 767 72.1 (67.4, 76.4) 1898 63.8 (59.8, 67.6) 1693 74.9 (71.3, 78.2) 2112 67.0 (63.3, 70.4) 0.32 0.39 93 69.9 (58.1, 79.6) 250 61.2 (57.1, 65.2) 186 72.0 (62.7, 79.8) 470 59.2 (53.7, 64.4) 2150 74.8 (71.6, 77.7) 3215 66.6 (62.8, 70.3) 0.53 0.034
225 65.6 (59.2, 71.6) 561 57.8 (50.7, 64.5) 2278 75.2 (71.8, 78.3) 3498 66.8 (63.2, 70.1) 0.002 0.020
259 67.6 (62.5, 72.3) 287 57.8 (50.0, 65.3) 1670 76.6 (73.0, 79.9) 1545 68.8 (64.9, 72.4) 514 69.6 (63.7, 75.0) 2182 63.9 (60.1, 67.5) b0.001 0.004 1769 76.4 (72.7, 79.6) 1234 71.8 (67.9, 75.4) 616 69.6 (64.2, 74.5) 2495 62.7 (58.7, 66.5) 106 64.2 (53.2, 73.9) b0.001
312 60.3 (53.8, 66.3) b0.001
605 68.8 (63.7, 73.4) 2367 62.2 (58.2, 66.1) 1643 76.7 (73.2, 79.8) 1140 71.9 (68.0, 75.6) 124 76.6 (65.5, 85.0) 89 69.7 (58.6, 78.8) 66 62.1 (51.5, 71.7) 86 75.6 (63.7, 84.5) 0.010
276 65.6 (58.2, 72.3) 223 65.0 (59.6, 70.1) 0.005
344 77.9 (73.2, 82.0) 304 65.0 (61.3, 68.6) 2085 73.6 (69.7, 77.2) 3610 69.1 (64.4, 35.6) 0.094 0.22
1223 74.6 (70.9, 78.1) 1883 65.8 (61.8, 69.7) 885 74.7 (70.0, 78.8) 1522 66.0 (61.5, 70.2) 308 73.4 (65.5, 80.0) 518 62.9 (58.5, 67.2) 95 69.5 (58.3, 78.7) 0.58 1499 727 184 95
76.9 (73.0, 70.3 (66.1, 74.5 (68.3, 61.0 (44.5, 0.036
133 61.6 (53.8, 69.0) 0.65
80.4) 925 72.0 (68.2, 74.1) 2474 63.3 (59.5, 79.8) 507 65.3 (59.6, 75.4) 141 59.6 (52.0, 0.001
75.5) 67.0) 70.5) 66.8)
numbers of GPs), the risk of vaccination of individuals in practices in the middle UPA tertile was no different to that from the top (least deprived) UPA tertile but patients in practices in the most deprived UPA tertile were less likely to have had the vaccine (RR = 0.88, 95% CI 0.80, 0.96). Within each UPA tertile, uptake in practices in the highest SMR tertile was 0.87 (95% CI 0.81, 0.95) that of the lowest. Similarly, uptake in patients in practices in the middle SMR tertile was 0.86 (95% CI 0.79, 0.94) that of the highest tertile.
Discussion In this study, in a population of over 74 year olds, 64% had been vaccinated against influenza in 2000. Those without central heating or in rented accommodation were, at the most, 10% less likely to be vaccinated than owneroccupiers with central heating and those in supported homes (sheltered or residential). These modest socioeconomic differentials were independent of density of population, practice location in terms of deprivation and mortality rate, gender, health status, social support, or smoking. The study was not designed to assess the relative importance of the practice location variables (UPA and SMR tertiles) and individual factors, because the former were stratifiers and not intended for modelling as risk factors. However, when they were jointly examined in models with personal socioeconomic position, they remained significant factors suggesting other factors related to practice location affect uptake. The UPA score, originally intended as a marker of demand on practice services, has been validated against practice consultation data and mortality outcomes that could be influenced by primary care treatment [10]. Practice factors such as size, number of GPs, use of active recruitment, and computerisation did not appear here to be reasons for the importance of practice location. The findings also suggested that other factors are of importance. Women have been shown to be less likely to be vaccinated than men [3] and non-married less than married. Notes to Table 4: a P value for heterogeneity from models adjusted for age and accounting for clustering and stratification. b At least one of heart attack, stroke, possible angina. c At least one of: shortness of breath; chronic cough or phlegm; wheeze; having ever been diagnosed with pneumonia, emphysema, or asthma. d At least one of: poor/fair self-reported general health compared to others of the same age; perceiving self as not very or not at all active; being in lowest quintile of BMI; unable to do two or more out of eight activities of daily living. e Score on the Geriatric Depression Scale. Higher score is worse. Cut off of 5 or more commonly used to identify clinical depression of any severity. f Score of less than 24/30 on the Mini-Mental State Examination (MMSE) if able to answer language questions or less than 17/21 if unable to answer most of the language section. g Someone can really count on or feel at ease with when want to talk about private matters or when worried or stressed.
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
551
Table 5 Risk ratios (95% Confidence intervals) for obtaining vaccination in the winter 2000 season by socioeconomic variables and rural–urban area Part 1 n = 5720
Model 1 Adjusted for age, gender, not for other factors in table
Model 2 Adjusted for age, gender, and for other factors in table
Model 3 Adjusted for age, gender, practice factorsa, and for other factors in table
Personal socioeconomic position Owner-occupied Central heating None Rented Central heating None Supported housing P value
1.00 0.91 (0.84, 0.98)
1.00 0.91 (0.85, 0.98)
1.00 0.91 (0.85, 0.98)
0.92 (0.87, 0.99) 0.90 (0.79, 1.02) 0.99 (0.93, 1.06) 0.12
0.95 (0.89, 1.01) 0.92 (0.82, 1.02) 1.02 (0.96, 1.09) 0.13
0.95 (0.90, 1.01) 0.93 (0.84, 1.02) 1.03 (0.96, 1.11) 0.11
Carstairs’ index Least deprived 2nd quintile Middle quintile Fourth quintile Most deprived P value
1.00 1.01 (0.93, 0.93 (0.86, 1.02 (0.92, 0.88 (0.76, 0.026
1.00 1.00 (0.93, 0.94 (0.87, 1.03 (0.93, 0.90 (0.78, 0.030
1.00 0.97 (0.91, 0.92 (0.86, 1.02 (0.94, 0.91 (0.79, 0.026
Urban indicator Rural 250–1000 1000–2500 Urban P value Part 2
1.09) 1.01) 1.13) 1.02)
1.07) 1.01) 1.15) 1.03)
1.03) 0.99) 1.11) 1.04)
1.00 0.94 (0.84, 1.06) 0.93 (0.82, 1.05) 1.13 (1.03, 1.24) b0.001
1.00 0.94 (0.84, 1.04) 0.93 (0.83, 1.05) 1.12 (1.02, 1.23) 0.003
1.00 0.94 (0.85, 1.03) 0.95 (0.86, 1.04) 1.10 (1.01, 1.21) 0.020
Model 4 As model 3 and adjusted for health factorsb
Model 5 As model 4 and adjusted smokingc
Model 6 As model 5 and adjusted for support factorsd
Personal socioeconomic position Owner-occupied Central heating None Rented Central heating None Supported housing P value
1.00 0.91 (0.85, 0.98)
1.00 0.92 (0.85, 0.99)
1.00 0.93 (0.87, 1.00)
0.96 (0.90, 1.02) 0.94 (0.85, 1.03) 1.04 (0.97, 1.12) 0.11
0.96 (0.90, 1.02) 0.94 (0.86, 1.03) 1.05 (0.98, 1.12) 0.10
0.97 (0.91, 1.03) 0.95 (0.87, 1.04) 1.06 (0.99, 1.14) 0.10
Carstairs’ index Least deprived 2nd quintile Middle quintile Fourth quintile Most deprived P value
1.00 0.97 (0.91, 0.92 (0.86, 1.02 (0.94, 0.92 (0.80, 0.038
1.00 0.97 (0.92, 0.93 (0.86, 1.03 (0.94, 0.93 (0.81, 0.034
1.00 0.97 (0.92, 0.93 (0.86, 1.03 (0.95, 0.93 (0.81, 0.034
Urban indicator Rural 250–1000 1000–2500 Urban P value
1.00 0.95 (0.86, 1.04) 0.96 (0.87, 1.05) 1.12 (1.02, 1.22) 0.016
a b c d
1.03) 0.99) 1.12) 1.06)
1.04) 0.99) 1.12) 1.07)
1.00 0.95 (0.87, 1.03) 0.95 (0.87, 1.05) 1.12 (1.02, 1.22) 0.014
Number of GPs in practice (less than 7; 7 or more); list size (less than 5000, 5000 or more). At least one respiratory problem, depression score, cognitive deficit. Current cigarette smoking (never/ex/current). Marital status (currently married or living with a partner/formerly married/single).
1.04) 0.99) 1.13) 1.07)
1.00 0.95 (0.87, 1.04) 0.96 (0.87, 1.05) 1.12 (1.03, 1.23) 0.012
552
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
Only some of the vulnerable groups had increased take-up: those with respiratory disease but not smokers or those with a history of CVD. Those with depression or cognitive impairment appeared to need greater support in accessing the service. These differences did not account for the socioeconomic differentials in influenza vaccine uptake. The sample is not a complete sample of people aged over 74 years in 2000 but of the survivors of a cohort who took part in a b75+Q health check at some time during the previous 6 years. It is likely that the sample analysed was healthier than the general population of people aged 75 years and over. They were also noted to be more likely to have influenza vaccination than those who were not assessed. The socioeconomic differentials in uptake in 2000 seen here are likely to be underestimates even if absolute vaccination levels are higher than one would expect in the whole population. Many of the findings seen here in over 74 year olds are consistent with uptake patterns seen in over 64 year olds; low uptake in smokers [11] and high uptake in those with an underlying respiratory conditions [12,13]. The slightly low influenza vaccination uptake amongst over 74 year olds who were not socioeconomically privileged is in contrast to the much lower uptake of preventive services seen in general in over 64 year olds, inverse to need [14,15]. In fact in Canada, no socioeconomic variations in influenza vaccine uptake, independent of the very strong influence of recommendation by a doctor or nurse, were seen [16]. The reason for these modest differentials of an important preventive health measure in elderly people can only be speculated upon here but it may have to do with the perception of influenza as an often mild illness and an ineffective vaccine given the large number of other influenza like infections that occur [17]. There is little information on contextual influences such as urban/rural or area deprivation on uptake of preventive services, apart from a Spanish study which, in contrast to findings here, which found that uptake was higher in low density compared to high population density settings [13]. Randomised controlled trials of active reminders [18], computer produced lists of eligible persons, and use of reminder cards in the Netherlands [19], USA [20], and Trent region in the UK [21] increase uptake. [22]. The fact that no effect of active recruitment on vaccine uptake was seen here is likely to be due to active recruitment being used to remedy low uptake. A number of reasons for the patterns of influenza vaccine uptake seen here are possible. Information on ability to get to their general practice would be useful to determine whether easier access to the GP in urban areas may be a factor. Amongst chronically ill people aged over 64 years in Iowa, there was a clear positive influence of education on vaccine uptake [23], though not in a Spanish study [13]. Socioeconomic position is multi-dimensional and may vary with context. Other influences may also be a sense of control and status [24] as well as trust in medical expertise [25].
Inefficient boffice systemsQ may also be behind the lower uptake in practices in more deprived locations. A recent randomised trial to improve the organisational systems to deliver infant vaccinations resulted in higher uptake in the US [26]. In the UK, before there was any age-based policy for influenza vaccination, there were considerable difficulties in actively recruiting patients, ordering the right quantity of vaccine [27], and poor communication, leaving patients unaware of their high-risk status or existence of the vaccine. In one study, over 80% of those with underlying illnesses took up influenza vaccination, offered mostly by personal recommendation, when patients consulted the doctor for another reason [21,28]. Lack of knowledge may still underlie some of the differentials in uptake. Adult influenza vaccination of people aged over 64 years has been introduced as policy to reduce acute morbidity and mortality. National influenza vaccine uptake rates for the more vulnerable elderly population aged over 74 years are not collated systematically. Nor are there any routine data on the prevalence of people’s vulnerability such as disability, chronic ill health, socioeconomic status, and their relationships with influenza vaccine receipt. We found that influenza vaccine uptake in 2000 in over 74 year olds was lower if they were in practices located in areas either with deprived populations or with relatively high mortality rates on top of deprivation levels. These findings support weighting of resources to more deprived areas [29] as a contribution to equity of health in older people. There are also indications that people whose personal socioeconomic circumstances are better than others are more likely to be vaccinated, although the differentials were modest. Of importance was the greater vulnerability from poor mental health and social isolation. Greater support for such patient groups to access influenza vaccination is needed. Finally, surveillance of trends in influenza vaccine uptake by socioeconomic groups is a feasible contribution to reducing inequalities in health [30]. Practice-level influenza vaccine uptake data are now routinely collected [2] and could be correlated with routinely available practice SMR and deprivation scores for monitoring, supplemented by practice-based and occasional national surveys.
Acknowledgments This work was funded by a grant from the Department of Health’s Inequalities in Health Research Initiative (ref no 121/7415). PM was partially funded by a Wellcome Trust grant (grant number 051637) during the work. We would like to thank Gill Price for statistical help. The authors would also like to thank the MRC Trial investigators and Trial Steering Committee of the MRC Trial for allowing them to use the Trial data. The investigators, other than Professor Astrid Fletcher, were Professor Chris Bulpitt (Imperial College London), Dr. Dee Jones (University of
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
Wales) and Dr. Alistair Tulloch (University of Oxford). Other members of the Steering Committee were Professor Sir John Grimley Evans (University of Oxford), Professor Carol Brayne (University of Cambridge), Professor Karen Luker (University of Manchester), Professor Mike Drummond (University of York). Data collection for the main Trial were undertaken by research nurses from The General Practice Research Framework under the guidance of its Director Dr. Madge Vickers and the Senior Research Nurse Nicola Fasey. Susan Stirling was the trial statistician until September 2000. Data management, clerical, and administrative support were provided by Maria Nunes and Ruth Peters (Imperial College, London), Susana Scott (until 1999), Edmond Ng (from 2000), Janbibi Mazar, and Rakhi Kabawala (LSHTM). Chris Grundy (LSHTM) undertook the postcode linkage.
References [1] Chief Medical Officer. Influenza immunisations. 1st August 2000. Department of Health PL/CMO/2000/3. Department of Health, 2000. [2] Joseph C, Goddard NL. Influenza vaccine uptake in the elderly: results from a rapid assessment of the effectiveness of new government policy in England for the winters 2000/2001 and 2001/2002. Vaccine 2003;21:1137 – 48. [3] Breeze E, Mangtani P, Fletcher AE, Price GM, Kovats S, Roberts JA, Trends in influenza vaccination uptake among people aged over 74 years, 1997–2000: survey of 73 General Practices in Britain. BMC Family Practice. 2004;90:160–6. [4] Fletcher AE, Jones DA, Bulpitt CJ, Tulloch AJ. The MRC trial of assessment and management of older people in the community: objectives, design and interventions. BMC Health Serv Res 2002;2:21 (http://www.biomedcentral.com/1472-6963/2/21). [5] Jarman B. Identification of underprivileged areas. BMJ 1983;286: 1705 – 9. [6] Carstairs V, Morris R. Deprivation and mortality: an alternative to social class? Community Med 1989;11:210 – 9. [7] Sheik JI, Yesavage JA. Geriatric Depression Scale (GDS): recent evidence and development of a shorter version. Clin Gerontol 1986;37:819 – 20. [8] Folstein MF, Folstein SE, McHugh PR. bMini-mental stateQ. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1973;12:189 – 98. [9] Stata Corporation PR. Intercooled Stata, vol. 7. Texas7 Stata Press; 2001. [10] Smith GD. Second thoughts on the Jarman index. BMJ 1991;302:359 – 60. [11] Watkins J. Effectiveness of influenza vaccination policy at targeting patients at high risk of complications during winter 1994–5: cross sectional survey. BMJ 1997;315:1069 – 70.
553
[12] Honkanen PO, Keistinen T, Kivela SL. Factors associated with influenza vaccination coverage among the elderly: role of health care personnel. Public Health 1996;110:163 – 8. [13] Sarria-Santamera A, Timoner J. Influenza vaccination in old adults in Spain. Eur J Public Health 2002;13:133 – 7. [14] McCormick A, Fleming DM, Charlton J. Morbidity statistics from General Practice. Fourth national study 1991–1992. London7 HMSO; 1995. [15] Lorant V, Boland B, Humblet P, Deliege D. Equity in prevention and health care. J Epidemiol Community Health 2002;56:510 – 6. [16] Duclos P, Hatcher J. Epidemiology of influenza vaccination in Canada. Can J Public Health 1993;84:311 – 5. [17] vanEssen GA, Kuyvenhoven MM, de Melker RA. Why do healthy elderly people fail to comply with influenza vaccination? Age Ageing 1997;26:275 – 9. [18] Szilagyi PG, Bordley C, Vann JC, Chelminski A, Kraus RM, Margolis PA, Rodewald LE. Effect of patient reminder/recall interventions on immunization rates: a review. JAMA 2000;284: 1820 – 7. [19] Perenboom RJM, Davidse W. Increasing the coverage of vaccination against influenza by general practitioners. J Public Health Med 1996;18:183 – 7. [20] Mullooly JP. Increasing influenza vaccination among high-risk elderly: randomized controlled trial of a mail cue in an HMO setting. Am J Public Health 1987;77:626 – 7. [21] Nicholson KG, Wiselka MJ, May A. Influenza vaccination of the elderly: perceptions and policies of general practitioners and the outcome of the 1985–86 immunisation in Trent, UK. Vaccine 1987;5:302 – 6. [22] Barton MB, Schoenbaum SC. Improving influenza vaccination performance in an HMO setting: the use of computer generated reminders and peer comparison feedback. Am J Public Health 1990;80:534 – 6. [23] Petersen RL, Saag K, Wallace RB, Doebbeling BN. Influenza and pneumococcal vaccine receipt in older perons with chronic disease. Med Care 1999;37:502 – 9. [24] S. Macintyre, A. Kearns, A. Ellaway. Housing tenure and car ownership: why do they predict health and longevity? Final report of ESRC grant L12851018. 1–17. ESRC, editor. (2000). [25] Telford R, Rogers A. What influences elderly peoples’ decisions about whether to accept the influenza vaccination? A qualitative study. Health Educ Res 2003;18:743 – 53. [26] Margolis PA, Lannon CM, Stuart JM, Fried BJ, Keyes-Elstein L, Moore DE. Practice based education to improve delivery systems for prevention in primary care: randomised trial. BMJ 2004;328: 388 – 400. [27] Wiselka M. Influenza: diagnosis, management and prophylaxis. BMJ 1994;308:1341 – 5. [28] Nguyen-Van-Tam JS, Nicholson KG. Influenza immunisation: vaccine offer, request and uptake in high risk patients during the 1991/2 season. Epidemiol Infect 1993;111:347 – 55. [29] Bajekal M, Alves B, Jarman B, Hurwitz B. Rationale for the new GP deprivation payment scheme in England: effects of moving from electoral ward to enumeration district underprivileged area scores. Br J Gen Pract 2001;467:677. [30] Macintyre S. Evidence based policy making. BMJ 2003;326:5 – 6.
Inequalities in influenza vaccine uptake among people aged over 74 years in Britain Punam Mangtani, M.Sc., M.D.T, Elizabeth Breeze, Ph.D., Sari Kovats, M.Sc., Edmond S.W. Ng, M.Sc., Jennifer A. Roberts, Ph.D., Astrid Fletcher, Ph.D. London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK Available online 1 April 2005
Abstract Background. In the UK, payments to providers (General Practitioners) for vaccinating all people aged over 64 years old against influenza commenced in 2000. Little information exists on the relationship between uptake and need. We assessed factors influencing uptake and equity in uptake in over 74 year olds. Methods. We analysed cohort data from 5572 subjects in a community-based trial. Analyses took into account clustering and location of general practices in terms of Underprivileged Area (UPA) Score and area Standardised Mortality Ratio (SMR). Results. Vaccine uptake in practices in the most deprived tertile was 0.88 (95% CI 0.80–0.96) that of the least deprived and mid tertile, adjusted for confounding. Within each deprivation tertile, uptake in the mid and highest SMR tertile was 0.86 (95% CI 0.79, 0.94) and 0.87 (95% CI 0.81, 0.95) that of the lowest respectively. Uptake was 10% lower at the most in individuals with poorer quality housing. Higher uptake if married or with respiratory conditions and lower uptake if smoked had cognitive impairment or depression did not explain the socioeconomic differentials. Conclusions. Lower uptake in practices in deprived areas supports targeting of resources. At the individual level, those who are more isolated require support to access influenza vaccination. D 2005 Published by Elsevier Inc. Keywords: Influenza vaccine; Vaccination utilisation; Socioeconomic factors; Elderly
Introduction In 2000, the influenza vaccination programme was extended to all those over 64 in the UK [1]. General practices in the UK, as the source of all ambulatory care, are the main providers of influenza vaccination. Funds were made available to identify and approach all those eligible, monitor uptake, and support general practices, with insufficient staff. General Practitioners also received an item of service payment for vaccinations. There was some concern that influenza vaccine delivery would be inversely related to need as with other preventive services. Information, however, on the equity of influenza vaccination receipt and factors that could be addressed to improve uptake amongst elderly people is limited. Uptake T Corresponding author. Fax: +44 207 636 8739. E-mail address: [email protected] (P. Mangtani). 0091-7435/$ - see front matter D 2005 Published by Elsevier Inc. doi:10.1016/j.ypmed.2005.02.001
amongst people aged 64 years and older was 61% nationally in 2000 [2]. An initial analysis in a study population of over 74 year olds indicated that uptake increased from 51% in 1999 to 64% in 2000, but with no significant change in the fact that those with greatest socioeconomic disadvantage were less likely to be vaccinated than others [3]. This study sought to determine whether wider factors, such as pre-existing ill health, depression, cognitive impairment, and social support influenced vaccine uptake and whether these accounted for differences by socioeconomic position. The vaccination services of practices, such as active recruitment, were also examined.
Methods We used data on patients aged over 74 years in the UK Medical Research Council (MRC) Trial of Assess-
546
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
ment and Management of Older People in the Community. The general practice was the unit of randomisation. The design and methods are described elsewhere [4]. The trial was conducted in 106 practices recruited through the MRC General Practice Research Framework (GPRF) and selected to be representative of the joint tertiles of the UPA area deprivation scores [5] and Standard Mortality Ratios (SMRs) for electoral wards in Britain. The tertiles of SMR were nested within the tertiles of the UPA score. The trial evaluated a package of multi-dimensional assessment and management of older people. One of the main components was the method of assessment—Universal or Targeted. People eligible for the bover 74Q health assessment, excluding those in nursing homes or terminally ill, were invited to participate. The analysis of influenza vaccine uptake was confined to patients in practices in the Universal Arm who were invited as only these patients had full information collected on underlying medical conditions and socioeconomic factors based on a detailed health and social assessment by a study nurse. The assessments took place between 1995 and 2000 depending on when the practice was recruited. Patients were followed up to assess if they still belonged to the practice in 2000 by flagging for deaths and embarkations at the National Health Service Central Register (NHSCR) and by practices notifying changes of GP. Data on individual flu vaccination status were requested from practices. They were also asked about organisational factors likely to affect vaccination rates including patient reminders, extent to which registers of serious diseases were held, and level of computerisation, i.e. whether consultations and information on underlying medical conditions were recorded. Patients included had to be registered with the practice on 30th November 2000 when 95% of vaccinations had been given. The trial patients did not include those in nursing homes. Housing tenure was classified into owner-occupation, rented, and supported accommodation (sheltered and residential home). Supported accommodation was treated separately because it does not necessarily reflect the previous socioeconomic position of the occupier. Central heating in the home was combined with housing tenure as a marker of socioeconomic position. As type of area may determine ease of access to vaccination, the Carstairs’ deprivation index for the Enumeration District (ED) in which the patient resided was used as a socioeconomic marker of the area [6]. Population density from census data (persons per square kilometre within 5 km radius of the centroid of the ED) was also used, categorised into: urban (over 2500 persons/km2); 1000–2500 persons/ km2; 250–999 persons/km2; and rural (b250 persons/km2). Apart from gender and age, the other potential variables considered were: (i) Practice factors: list size, number of General Practioners (GPs), UPA and SMR tertiles of the ward in
which the practice was located, previous fund-holder status (where some payments for secondary care was directed by practices as opposed to centrally), computerisation level, and practice recruitment policy. (ii) Composite physical health indicators as reported by patients: respiratory history (at least one of shortness of breath, chronic cough or phlegm indicating chronic bronchitis, wheeze, ever-diagnosed by a doctor as having pneumonia or emphysema or asthma); cardiovascular history (at least one of heart attack or stroke as diagnosed by a doctor or a history of possible angina); general frailty (at least one of poor or fair self-reported health compared to others of the same age, perceiving oneself as not very or not at all active, being in the lowest quintile of BMI, or unable to do two or more out of eight activities of daily living). (iii) Mental health or ability. Depression was assessed on the Geriatric Depression Scale [7]. The Mini Mental State Examination [8] was used to suggest some cognitive deficit. (iv) Current smoking status. (v) Social support, such as marital status/living with a partner, contact with people outside the household, and presence of a confidante. All analyses took into account the clustered nature of the sample by using robust estimates derived from the bsvyQ suite of commands available in Stata 7 software [9]. Multilevel modelling was not used because the average number of EDs per practice was small, many EDs only contained one person, and the interest was in average effects. Modelling of the outcome relative to socioeconomic position and the potential for explanation was carried out using Poisson regression without a time factor to obtain risk ratios for being vaccinated or not. Unless otherwise specified, analyses included UPA score and SMR tertiles as stratifiers. All models were adjusted for gender and age. First, models were constructed for groups of explanatory factors to see which were predictors. The significant component factors in these groups were then added into models in a sequence felt to reflect the pathways of influence on vaccination (Fig. 1). In order to see whether a longer time lapse between measurement of these factors and autumn 2000 could lead to a dilution of effect, interactions between time lapse and the socioeconomic factors were tested. We examined for whether the effects of socioeconomic position differed by gender or underlying health.
Results Response rate Valid data for 2000 were obtained from 64% (34/53) practices in the Universal Arm of the trial. Participating
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
547
Fig. 1. Framework for models, indicating the stage at which explanatory factors were added.
practices compared to all eligible practices were more likely to be previous fund-holders, to have a large list size, or to be in the best UPA and SMR tertile (Table 1). The analyses of socioeconomic and other factors for uptake were carried out on 5572 people with complete information from the nurse assessment on all factors. The derivation of this subgroup for analysis is given in Fig. 2. Vaccine uptake amongst all those with information on vaccine status (including people who had not participated in the nurse assessment) (n = 9152) was 64% (95% CI 60– 67) (for sex specific crude uptake, see Table 2). People who participated in the nurse assessment (n = 6643), adjusted for age and gender, were 1.3 (95% CI 1.2, 1.4) times more likely to receive a vaccination than non-participants (n = 2509).
Vaccine uptake in relation to practice and individual characteristics At the crude level, people in practices with either seven or more GPs or less than 5000 list size were less likely to be vaccinated than those in other practices. Recruitment policy and level of computerisation made no substantial difference to uptake (Table 2). Vaccination uptake, adjusted for age in regression models, was highest amongst people in owner-occupation with central heating and amongst those in urban areas (unadjusted vaccination percentages are given in Table 3). The pattern of uptake by Carstairs’ quintile was irregular but with some indication of lower uptake in the most deprived areas. Preliminary tabulations by gender, adjusted for age (Table 4), suggested that the groups least likely to be
548
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
Table 1 Ratio of practices participating in 2000 to all practices in the Universal Arm by practice characteristics All
Fund-holding
34/53
Yes No Scottish
List size 11/16 20/32 3/5
b5000 7499 9999 10,000+
No. GPs 11/19 11/17 5/7 7/10
vaccinated were those: without respiratory problems, some cognitive deficit, a smoker, not married, living alone or with son/daughter, not having a confidante, and some indication of depression for women. Vaccine uptake in male smokers was lower compared to non smokers. Neither a history of cardiovascular disease (CVD) nor general frailty was significant predictors. Although the factors associated with vaccine uptake were not identical for males and females, separate models for each gender showed that the final models were very similar so results are shown here with men and women combined. All the individual predictors also had an association with at least one of the socioeconomic measures (not shown). Factors that were independently statistically associated with vaccination within each group shown in Fig. 1 (adjusted for gender and age and allowing for clustering and stratification by practice UPA and SMR tertile) were retained for modelling of socioeconomic differentials. These were: number of GPs and list size, respiratory vulnerability, depression, cognitive deficit, smoking status, and marital status. The general pattern for personal socioeconomic position (as measured
2 3 4–6 7+
UPA tertile 10/14 9/14 10/18 5/7
Best Middle Worst
SMR tertile 13/17 9/16 12/20
Best Middle Worst
11/16 12/17 11/20
by housing tenure and central heating) as described below was unchanged by adjustment for UPA and SMR.
Modelling socioeconomic position and vaccine uptake Table 5 shows the results for socioeconomic/geographic factors of the cumulative models. Examining each socioeconomic factor separately suggested that vaccine uptake was highest amongst people in owner-occupied homes or in supported housing and in the most urban area (model 1). While there was variation in uptake across Carstairs’ deprivation quartiles, there was no clear trend. Controlling for Carstairs’ score, level of urbanisation, health status, and social support did not alter the general pattern of uptake related to personal socioeconomic position (models 2–6). Although the overall P value is only of the order of 0.10, the risk ratio estimates for people in private households other than owner-occupiers with central heating are all below 1.0 and similar to each other suggesting that collectively they are at modest disadvantage.
Fig. 2. Analysis sample of vaccination in 2000 compared with fuller Trial sample.
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
549
Table 2 Percentages of men and women vaccinated in the winter 2000 season by practice factors Practice factors
Men
All List size b5000 5000 or more Recruitment policy Active Passive Number of GPs 2–6 7+ Previous fund-holding status Fund-holder Not a fund-holder Scottish practice Computerisation of consultations Mostly manual Mostly computerised All computerised and some underlying conditions All computerised and all underlying conditions a
Women a
n
% (95% CI)
n
% (95% CI)a
3293
69.8% (66.5–72.8)
5859
60.5% (56.9, 64.1)
1759 410
71.8 (67.9, 75.4) 82.7 (77.9, 86.6) P = 0.001
2983 568
63.3 (59.3, 67.1) 72.2 (65.5, 78.0) P = 0.020
1122 676
75.2 (70.6, 79.2) 71.0 (64.3, 76.9) P = 0.36
2455 1096
65.3 (60.2, 70.0) 63.5 (58.7, 68.0) P = 0.67
1741 428
75.1 (71.3, 78.6) 68.7 (64.8, 72.3) P = 0.014
2791 760
66.5 (62.9, 69.9) 58.2 (53.6, 62.6) P = 0.002
826 1172 171
73.1 (67.4, 78.1) 75.1 (70.0, 79.6) 69.0 (60.8, 76.2) P = 0.41
1321 1884 346
63.3 (58.5, 67.9) 67.0 (62.1, 71.5) 57.8 (53.3, 62,2) P = 0.014
70.1 77.2 66.3 72.8
382 1487 419 1129
62.0 65.6 56.3 66.6
224 900 252 716
(62.5, (71.4, (54.3, (66.7,
76.7) 82.2) 76.4) 78.1) P = 0.41
(55.5, (58.4, (49.2, (61.4,
68.1) 72.1) 63.2) 71.4) P = 0.20
P values for practice factors from model adjusted for age.
The risk ratio for being vaccinated if in owner occupation without heating compared to those with central heating was 0.90 (0.85, 0.95) for people assessed 3 or more years previously and 1.14 (0.98, 1.33) for those more recently assessed, suggesting that there was no dilution of socio-
differentials with time since assessment. There were also no interactions between either gender or frailty and socioeconomic status (data not presented). After adjustment for personal socioeconomic position and other significant practice risk factors (list size and
Table 3 Percentages of men and women vaccinated in the winter 2000 season by personal socioeconomic position and geographic factors Men
Women
n
% (95% CI)a
n
% (95% CI)a
1710 231
76.0 (72.2, 79.4) 69.7 (61.9, 76.5)
2427 380
68.2 (64.4, 71.7) 58.7 (51.9, 65.2)
401 75 89
70.6 (65.3, 75.4) 64.0 (53.4, 73.4) 77.5 (68.9, 84.3) 0.089
744 139 364
62.8 (58.5, 66.8) 61.9 (51.4, 71.3) 62.4 (57.5, 67.0) 0.049
Carstairs’ index quintiles b Least deprived Second quintile Third quintile Fourth quintile Most deprived P value
729 666 531 288 133
73.1 (67.8, 76.7 (71.3, 72.3 (67.3, 77.8 (70.4, 63.9 (54.9, 0.078
77.8) 81.4) 76.8) 83.8) 72.1)
1081 1094 840 492 269
67.1 (62.4, 65.9 (60.8, 60.7 (56.2, 66.5 (58.6, 59.5 (51.0, 0.125
71.4) 70.7) 65.1) 73.5) 67.4)
Urban indicator c pers pkmsq Rural 250–1000 1000–2500 Urban P value
739 611 602 395
74.4 (68.0, 71.2 (63.4, 70.9 (65.0, 82.3 (76.5, 0.077
79.9) 77.9) 76.2) 86.9)
1183 1001 976 616
64.7 (58.7, 62.0 (56.2, 60.7 (54.8, 75.5 (71.1, 0.005
70.2) 67.5) 66.2) 79.4)
Personal socioeconomic position Owner-occupier Central heating None Rent Central heating None Supported P value
a b c
P value for heterogeneity from models adjusted for age; confidence intervals adjusted for sample design. Quintiles derived from distribution of Carstairs’ scores for EDs throughout Britain. Based on population density within 5 km radius of centroid of ED.
550
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
Table 4 Percentages of men and women vaccinated in the winter 2000 season by health status, smoking, and support Men b
History of cvd None P value History of respiratory problemsc None P value Indicator frailtyd Not P value Depression scoree More than 5 4–5 Less than 4 P value Cognitive deficit (MMSE)f Some None P value Cigarette smoking status Current Ex Never P value Marital status Married/co-habiting Widowed/divorced/ separated Single P value Living status Alone With spouse only With spouse and other With son/daughter With other P value Carer for someone else at home Yes No P value Social contact outside household Daily 2–3 times a week Less than twice a week Rarely P value Has confidanteg Spouse (and other) Other relative Other No confidante P value
Women % (95% CI)a
n
n
% (95% CI)a
579 77.6 (72.8, 81.7) 696 66.2 (60.7, 71,4) 1938 73.4 (69.6, 76.8) 3381 65.3 (61.9, 68.5) 0.082 0.451 1368 76.3 (71.9, 80.2) 2061 67.4 (63.9, 70.8) 1088 71.5 (68.9, 74.2) 1853 63.5 (59.3, 67.50) 0.014 0.006 767 72.1 (67.4, 76.4) 1898 63.8 (59.8, 67.6) 1693 74.9 (71.3, 78.2) 2112 67.0 (63.3, 70.4) 0.32 0.39 93 69.9 (58.1, 79.6) 250 61.2 (57.1, 65.2) 186 72.0 (62.7, 79.8) 470 59.2 (53.7, 64.4) 2150 74.8 (71.6, 77.7) 3215 66.6 (62.8, 70.3) 0.53 0.034
225 65.6 (59.2, 71.6) 561 57.8 (50.7, 64.5) 2278 75.2 (71.8, 78.3) 3498 66.8 (63.2, 70.1) 0.002 0.020
259 67.6 (62.5, 72.3) 287 57.8 (50.0, 65.3) 1670 76.6 (73.0, 79.9) 1545 68.8 (64.9, 72.4) 514 69.6 (63.7, 75.0) 2182 63.9 (60.1, 67.5) b0.001 0.004 1769 76.4 (72.7, 79.6) 1234 71.8 (67.9, 75.4) 616 69.6 (64.2, 74.5) 2495 62.7 (58.7, 66.5) 106 64.2 (53.2, 73.9) b0.001
312 60.3 (53.8, 66.3) b0.001
605 68.8 (63.7, 73.4) 2367 62.2 (58.2, 66.1) 1643 76.7 (73.2, 79.8) 1140 71.9 (68.0, 75.6) 124 76.6 (65.5, 85.0) 89 69.7 (58.6, 78.8) 66 62.1 (51.5, 71.7) 86 75.6 (63.7, 84.5) 0.010
276 65.6 (58.2, 72.3) 223 65.0 (59.6, 70.1) 0.005
344 77.9 (73.2, 82.0) 304 65.0 (61.3, 68.6) 2085 73.6 (69.7, 77.2) 3610 69.1 (64.4, 35.6) 0.094 0.22
1223 74.6 (70.9, 78.1) 1883 65.8 (61.8, 69.7) 885 74.7 (70.0, 78.8) 1522 66.0 (61.5, 70.2) 308 73.4 (65.5, 80.0) 518 62.9 (58.5, 67.2) 95 69.5 (58.3, 78.7) 0.58 1499 727 184 95
76.9 (73.0, 70.3 (66.1, 74.5 (68.3, 61.0 (44.5, 0.036
133 61.6 (53.8, 69.0) 0.65
80.4) 925 72.0 (68.2, 74.1) 2474 63.3 (59.5, 79.8) 507 65.3 (59.6, 75.4) 141 59.6 (52.0, 0.001
75.5) 67.0) 70.5) 66.8)
numbers of GPs), the risk of vaccination of individuals in practices in the middle UPA tertile was no different to that from the top (least deprived) UPA tertile but patients in practices in the most deprived UPA tertile were less likely to have had the vaccine (RR = 0.88, 95% CI 0.80, 0.96). Within each UPA tertile, uptake in practices in the highest SMR tertile was 0.87 (95% CI 0.81, 0.95) that of the lowest. Similarly, uptake in patients in practices in the middle SMR tertile was 0.86 (95% CI 0.79, 0.94) that of the highest tertile.
Discussion In this study, in a population of over 74 year olds, 64% had been vaccinated against influenza in 2000. Those without central heating or in rented accommodation were, at the most, 10% less likely to be vaccinated than owneroccupiers with central heating and those in supported homes (sheltered or residential). These modest socioeconomic differentials were independent of density of population, practice location in terms of deprivation and mortality rate, gender, health status, social support, or smoking. The study was not designed to assess the relative importance of the practice location variables (UPA and SMR tertiles) and individual factors, because the former were stratifiers and not intended for modelling as risk factors. However, when they were jointly examined in models with personal socioeconomic position, they remained significant factors suggesting other factors related to practice location affect uptake. The UPA score, originally intended as a marker of demand on practice services, has been validated against practice consultation data and mortality outcomes that could be influenced by primary care treatment [10]. Practice factors such as size, number of GPs, use of active recruitment, and computerisation did not appear here to be reasons for the importance of practice location. The findings also suggested that other factors are of importance. Women have been shown to be less likely to be vaccinated than men [3] and non-married less than married. Notes to Table 4: a P value for heterogeneity from models adjusted for age and accounting for clustering and stratification. b At least one of heart attack, stroke, possible angina. c At least one of: shortness of breath; chronic cough or phlegm; wheeze; having ever been diagnosed with pneumonia, emphysema, or asthma. d At least one of: poor/fair self-reported general health compared to others of the same age; perceiving self as not very or not at all active; being in lowest quintile of BMI; unable to do two or more out of eight activities of daily living. e Score on the Geriatric Depression Scale. Higher score is worse. Cut off of 5 or more commonly used to identify clinical depression of any severity. f Score of less than 24/30 on the Mini-Mental State Examination (MMSE) if able to answer language questions or less than 17/21 if unable to answer most of the language section. g Someone can really count on or feel at ease with when want to talk about private matters or when worried or stressed.
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
551
Table 5 Risk ratios (95% Confidence intervals) for obtaining vaccination in the winter 2000 season by socioeconomic variables and rural–urban area Part 1 n = 5720
Model 1 Adjusted for age, gender, not for other factors in table
Model 2 Adjusted for age, gender, and for other factors in table
Model 3 Adjusted for age, gender, practice factorsa, and for other factors in table
Personal socioeconomic position Owner-occupied Central heating None Rented Central heating None Supported housing P value
1.00 0.91 (0.84, 0.98)
1.00 0.91 (0.85, 0.98)
1.00 0.91 (0.85, 0.98)
0.92 (0.87, 0.99) 0.90 (0.79, 1.02) 0.99 (0.93, 1.06) 0.12
0.95 (0.89, 1.01) 0.92 (0.82, 1.02) 1.02 (0.96, 1.09) 0.13
0.95 (0.90, 1.01) 0.93 (0.84, 1.02) 1.03 (0.96, 1.11) 0.11
Carstairs’ index Least deprived 2nd quintile Middle quintile Fourth quintile Most deprived P value
1.00 1.01 (0.93, 0.93 (0.86, 1.02 (0.92, 0.88 (0.76, 0.026
1.00 1.00 (0.93, 0.94 (0.87, 1.03 (0.93, 0.90 (0.78, 0.030
1.00 0.97 (0.91, 0.92 (0.86, 1.02 (0.94, 0.91 (0.79, 0.026
Urban indicator Rural 250–1000 1000–2500 Urban P value Part 2
1.09) 1.01) 1.13) 1.02)
1.07) 1.01) 1.15) 1.03)
1.03) 0.99) 1.11) 1.04)
1.00 0.94 (0.84, 1.06) 0.93 (0.82, 1.05) 1.13 (1.03, 1.24) b0.001
1.00 0.94 (0.84, 1.04) 0.93 (0.83, 1.05) 1.12 (1.02, 1.23) 0.003
1.00 0.94 (0.85, 1.03) 0.95 (0.86, 1.04) 1.10 (1.01, 1.21) 0.020
Model 4 As model 3 and adjusted for health factorsb
Model 5 As model 4 and adjusted smokingc
Model 6 As model 5 and adjusted for support factorsd
Personal socioeconomic position Owner-occupied Central heating None Rented Central heating None Supported housing P value
1.00 0.91 (0.85, 0.98)
1.00 0.92 (0.85, 0.99)
1.00 0.93 (0.87, 1.00)
0.96 (0.90, 1.02) 0.94 (0.85, 1.03) 1.04 (0.97, 1.12) 0.11
0.96 (0.90, 1.02) 0.94 (0.86, 1.03) 1.05 (0.98, 1.12) 0.10
0.97 (0.91, 1.03) 0.95 (0.87, 1.04) 1.06 (0.99, 1.14) 0.10
Carstairs’ index Least deprived 2nd quintile Middle quintile Fourth quintile Most deprived P value
1.00 0.97 (0.91, 0.92 (0.86, 1.02 (0.94, 0.92 (0.80, 0.038
1.00 0.97 (0.92, 0.93 (0.86, 1.03 (0.94, 0.93 (0.81, 0.034
1.00 0.97 (0.92, 0.93 (0.86, 1.03 (0.95, 0.93 (0.81, 0.034
Urban indicator Rural 250–1000 1000–2500 Urban P value
1.00 0.95 (0.86, 1.04) 0.96 (0.87, 1.05) 1.12 (1.02, 1.22) 0.016
a b c d
1.03) 0.99) 1.12) 1.06)
1.04) 0.99) 1.12) 1.07)
1.00 0.95 (0.87, 1.03) 0.95 (0.87, 1.05) 1.12 (1.02, 1.22) 0.014
Number of GPs in practice (less than 7; 7 or more); list size (less than 5000, 5000 or more). At least one respiratory problem, depression score, cognitive deficit. Current cigarette smoking (never/ex/current). Marital status (currently married or living with a partner/formerly married/single).
1.04) 0.99) 1.13) 1.07)
1.00 0.95 (0.87, 1.04) 0.96 (0.87, 1.05) 1.12 (1.03, 1.23) 0.012
552
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
Only some of the vulnerable groups had increased take-up: those with respiratory disease but not smokers or those with a history of CVD. Those with depression or cognitive impairment appeared to need greater support in accessing the service. These differences did not account for the socioeconomic differentials in influenza vaccine uptake. The sample is not a complete sample of people aged over 74 years in 2000 but of the survivors of a cohort who took part in a b75+Q health check at some time during the previous 6 years. It is likely that the sample analysed was healthier than the general population of people aged 75 years and over. They were also noted to be more likely to have influenza vaccination than those who were not assessed. The socioeconomic differentials in uptake in 2000 seen here are likely to be underestimates even if absolute vaccination levels are higher than one would expect in the whole population. Many of the findings seen here in over 74 year olds are consistent with uptake patterns seen in over 64 year olds; low uptake in smokers [11] and high uptake in those with an underlying respiratory conditions [12,13]. The slightly low influenza vaccination uptake amongst over 74 year olds who were not socioeconomically privileged is in contrast to the much lower uptake of preventive services seen in general in over 64 year olds, inverse to need [14,15]. In fact in Canada, no socioeconomic variations in influenza vaccine uptake, independent of the very strong influence of recommendation by a doctor or nurse, were seen [16]. The reason for these modest differentials of an important preventive health measure in elderly people can only be speculated upon here but it may have to do with the perception of influenza as an often mild illness and an ineffective vaccine given the large number of other influenza like infections that occur [17]. There is little information on contextual influences such as urban/rural or area deprivation on uptake of preventive services, apart from a Spanish study which, in contrast to findings here, which found that uptake was higher in low density compared to high population density settings [13]. Randomised controlled trials of active reminders [18], computer produced lists of eligible persons, and use of reminder cards in the Netherlands [19], USA [20], and Trent region in the UK [21] increase uptake. [22]. The fact that no effect of active recruitment on vaccine uptake was seen here is likely to be due to active recruitment being used to remedy low uptake. A number of reasons for the patterns of influenza vaccine uptake seen here are possible. Information on ability to get to their general practice would be useful to determine whether easier access to the GP in urban areas may be a factor. Amongst chronically ill people aged over 64 years in Iowa, there was a clear positive influence of education on vaccine uptake [23], though not in a Spanish study [13]. Socioeconomic position is multi-dimensional and may vary with context. Other influences may also be a sense of control and status [24] as well as trust in medical expertise [25].
Inefficient boffice systemsQ may also be behind the lower uptake in practices in more deprived locations. A recent randomised trial to improve the organisational systems to deliver infant vaccinations resulted in higher uptake in the US [26]. In the UK, before there was any age-based policy for influenza vaccination, there were considerable difficulties in actively recruiting patients, ordering the right quantity of vaccine [27], and poor communication, leaving patients unaware of their high-risk status or existence of the vaccine. In one study, over 80% of those with underlying illnesses took up influenza vaccination, offered mostly by personal recommendation, when patients consulted the doctor for another reason [21,28]. Lack of knowledge may still underlie some of the differentials in uptake. Adult influenza vaccination of people aged over 64 years has been introduced as policy to reduce acute morbidity and mortality. National influenza vaccine uptake rates for the more vulnerable elderly population aged over 74 years are not collated systematically. Nor are there any routine data on the prevalence of people’s vulnerability such as disability, chronic ill health, socioeconomic status, and their relationships with influenza vaccine receipt. We found that influenza vaccine uptake in 2000 in over 74 year olds was lower if they were in practices located in areas either with deprived populations or with relatively high mortality rates on top of deprivation levels. These findings support weighting of resources to more deprived areas [29] as a contribution to equity of health in older people. There are also indications that people whose personal socioeconomic circumstances are better than others are more likely to be vaccinated, although the differentials were modest. Of importance was the greater vulnerability from poor mental health and social isolation. Greater support for such patient groups to access influenza vaccination is needed. Finally, surveillance of trends in influenza vaccine uptake by socioeconomic groups is a feasible contribution to reducing inequalities in health [30]. Practice-level influenza vaccine uptake data are now routinely collected [2] and could be correlated with routinely available practice SMR and deprivation scores for monitoring, supplemented by practice-based and occasional national surveys.
Acknowledgments This work was funded by a grant from the Department of Health’s Inequalities in Health Research Initiative (ref no 121/7415). PM was partially funded by a Wellcome Trust grant (grant number 051637) during the work. We would like to thank Gill Price for statistical help. The authors would also like to thank the MRC Trial investigators and Trial Steering Committee of the MRC Trial for allowing them to use the Trial data. The investigators, other than Professor Astrid Fletcher, were Professor Chris Bulpitt (Imperial College London), Dr. Dee Jones (University of
P. Mangtani et al. / Preventive Medicine 41 (2005) 545–553
Wales) and Dr. Alistair Tulloch (University of Oxford). Other members of the Steering Committee were Professor Sir John Grimley Evans (University of Oxford), Professor Carol Brayne (University of Cambridge), Professor Karen Luker (University of Manchester), Professor Mike Drummond (University of York). Data collection for the main Trial were undertaken by research nurses from The General Practice Research Framework under the guidance of its Director Dr. Madge Vickers and the Senior Research Nurse Nicola Fasey. Susan Stirling was the trial statistician until September 2000. Data management, clerical, and administrative support were provided by Maria Nunes and Ruth Peters (Imperial College, London), Susana Scott (until 1999), Edmond Ng (from 2000), Janbibi Mazar, and Rakhi Kabawala (LSHTM). Chris Grundy (LSHTM) undertook the postcode linkage.
References [1] Chief Medical Officer. Influenza immunisations. 1st August 2000. Department of Health PL/CMO/2000/3. Department of Health, 2000. [2] Joseph C, Goddard NL. Influenza vaccine uptake in the elderly: results from a rapid assessment of the effectiveness of new government policy in England for the winters 2000/2001 and 2001/2002. Vaccine 2003;21:1137 – 48. [3] Breeze E, Mangtani P, Fletcher AE, Price GM, Kovats S, Roberts JA, Trends in influenza vaccination uptake among people aged over 74 years, 1997–2000: survey of 73 General Practices in Britain. BMC Family Practice. 2004;90:160–6. [4] Fletcher AE, Jones DA, Bulpitt CJ, Tulloch AJ. The MRC trial of assessment and management of older people in the community: objectives, design and interventions. BMC Health Serv Res 2002;2:21 (http://www.biomedcentral.com/1472-6963/2/21). [5] Jarman B. Identification of underprivileged areas. BMJ 1983;286: 1705 – 9. [6] Carstairs V, Morris R. Deprivation and mortality: an alternative to social class? Community Med 1989;11:210 – 9. [7] Sheik JI, Yesavage JA. Geriatric Depression Scale (GDS): recent evidence and development of a shorter version. Clin Gerontol 1986;37:819 – 20. [8] Folstein MF, Folstein SE, McHugh PR. bMini-mental stateQ. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1973;12:189 – 98. [9] Stata Corporation PR. Intercooled Stata, vol. 7. Texas7 Stata Press; 2001. [10] Smith GD. Second thoughts on the Jarman index. BMJ 1991;302:359 – 60. [11] Watkins J. Effectiveness of influenza vaccination policy at targeting patients at high risk of complications during winter 1994–5: cross sectional survey. BMJ 1997;315:1069 – 70.
553
[12] Honkanen PO, Keistinen T, Kivela SL. Factors associated with influenza vaccination coverage among the elderly: role of health care personnel. Public Health 1996;110:163 – 8. [13] Sarria-Santamera A, Timoner J. Influenza vaccination in old adults in Spain. Eur J Public Health 2002;13:133 – 7. [14] McCormick A, Fleming DM, Charlton J. Morbidity statistics from General Practice. Fourth national study 1991–1992. London7 HMSO; 1995. [15] Lorant V, Boland B, Humblet P, Deliege D. Equity in prevention and health care. J Epidemiol Community Health 2002;56:510 – 6. [16] Duclos P, Hatcher J. Epidemiology of influenza vaccination in Canada. Can J Public Health 1993;84:311 – 5. [17] vanEssen GA, Kuyvenhoven MM, de Melker RA. Why do healthy elderly people fail to comply with influenza vaccination? Age Ageing 1997;26:275 – 9. [18] Szilagyi PG, Bordley C, Vann JC, Chelminski A, Kraus RM, Margolis PA, Rodewald LE. Effect of patient reminder/recall interventions on immunization rates: a review. JAMA 2000;284: 1820 – 7. [19] Perenboom RJM, Davidse W. Increasing the coverage of vaccination against influenza by general practitioners. J Public Health Med 1996;18:183 – 7. [20] Mullooly JP. Increasing influenza vaccination among high-risk elderly: randomized controlled trial of a mail cue in an HMO setting. Am J Public Health 1987;77:626 – 7. [21] Nicholson KG, Wiselka MJ, May A. Influenza vaccination of the elderly: perceptions and policies of general practitioners and the outcome of the 1985–86 immunisation in Trent, UK. Vaccine 1987;5:302 – 6. [22] Barton MB, Schoenbaum SC. Improving influenza vaccination performance in an HMO setting: the use of computer generated reminders and peer comparison feedback. Am J Public Health 1990;80:534 – 6. [23] Petersen RL, Saag K, Wallace RB, Doebbeling BN. Influenza and pneumococcal vaccine receipt in older perons with chronic disease. Med Care 1999;37:502 – 9. [24] S. Macintyre, A. Kearns, A. Ellaway. Housing tenure and car ownership: why do they predict health and longevity? Final report of ESRC grant L12851018. 1–17. ESRC, editor. (2000). [25] Telford R, Rogers A. What influences elderly peoples’ decisions about whether to accept the influenza vaccination? A qualitative study. Health Educ Res 2003;18:743 – 53. [26] Margolis PA, Lannon CM, Stuart JM, Fried BJ, Keyes-Elstein L, Moore DE. Practice based education to improve delivery systems for prevention in primary care: randomised trial. BMJ 2004;328: 388 – 400. [27] Wiselka M. Influenza: diagnosis, management and prophylaxis. BMJ 1994;308:1341 – 5. [28] Nguyen-Van-Tam JS, Nicholson KG. Influenza immunisation: vaccine offer, request and uptake in high risk patients during the 1991/2 season. Epidemiol Infect 1993;111:347 – 55. [29] Bajekal M, Alves B, Jarman B, Hurwitz B. Rationale for the new GP deprivation payment scheme in England: effects of moving from electoral ward to enumeration district underprivileged area scores. Br J Gen Pract 2001;467:677. [30] Macintyre S. Evidence based policy making. BMJ 2003;326:5 – 6.