Factors associated with pneumococcal polysaccharide vaccination of the elderly in Spain: A cross-sectional study

ABSTRACT

Vaccination of the elderly is an important factor in limiting the impact of pneumonia in the community. The aim of this study was to investigate the factors associated with pneumococcal polysaccharide vaccination in patients aged ≥ 65 years hospitalized for causes unrelated to pneumonia, acute respiratory disease, or influenza-like illness in Spain. We made a cross-sectional study during 2013-2014. A bivariate analysis was performed comparing vaccinated and unvaccinated patients, taking into account sociodemographic variables and risk medical conditions. A multivariate analysis was performed using multilevel regression models. 921 patients were included; 403 (43.8%) had received the pneumococcal vaccine (394 received the polysaccharide vaccine). Visiting the general practitioner ≥ 3 times during the last year (OR = 1.79; 95% CI 1.25-2.57); having received the influenza vaccination in the 2013-14 season (OR = 2.57; 95% CI 1.72-3.84) or in any of the 3 previous seasons (OR = 11.70; 95% CI 7.42-18.45) were associated with receiving the pneumococcal polysaccharide vaccine. Pneumococcal vaccination coverage of hospitalized elderly people is low. The elderly need to be targeted about pneumococcal vaccination and activities that encourage healthcare workers to proactively propose vaccination might be useful. Educational campaigns aimed at the elderly could also help to increase vaccination coverages and reduce the burden of pneumococcal disease in the community.

KEYWORDS:

• elderly patients
• immunization
• pneumococcal vaccination
• public health
• vaccination

Introduction

Streptococcus pneumoniae, an asymptomatic colonizer of the nasopharynx, is a major cause of infections including pneumonia, meningitis, bacteremia, sinusitis, and otitis media. Most studies of the etiology of hospitalized community-acquired pneumonia in adults rank S. pneumoniae first among all known causes.

Despite advances in antimicrobial therapy, early mortality due to pneumococcal bacteremia has remained constant at 5% to 10% over the last century.¹

Adults aged ≥65 years comprise about 15% of the population but have a high case-fatality rate (≥15%). In this age group, most invasive cases result from the complications of pneumonia, but 5–10 times as many older adults have pneumococcal pneumonia without bacteremia. Mortality increases substantially with age and is 2-to- 5-fold higher in adults with underlying diseases than in healthier older adults.²

Recognition of continued morbidity and mortality from pneumococcal infections despite the use of appropriate antibiotics led to increased interest in disease prevention by vaccination and, since 1983, a 23-valent pneumococcal polysaccharide vaccine (PPSV) containing antigens against 23 of the 94 serotypes has been available.¹ Post-licensure studies show the vaccine is protective against invasive pneumococcal disease (IPD) in immunocompetent older adults.^3-5 Evidence supporting a beneficial effect of PPSV in preventing pneumococcal pneumonia is more limited, but some studies have shown benefits.^6-9 The median hospital stay in hospitalized adults with community-acquired pneumococcal pneumonia has been shown to be shorter in vaccinated patients¹⁰ and it has been reported that, in elderly people with chronic illness, the PPSV may reduce hospitalization during the influenza season,⁶ while, in people vaccinated with both the PPSV and the influenza vaccine, the benefit against hospitalization was greater than in those receiving only the influenza vaccine.^11,12 The World Health Organization states that the PPSV has demonstrated a protective effect against IPD and all-cause pneumonia among healthy young adults and, to a lesser extent, against IPD in elderly people who are not severely immunosuppressed.¹³ Cost-effectiveness studies support the recommendation of pneumococcal vaccination of the elderly, especially when the expected increase in the population aged ≥ 65 y in forthcoming years is taken into account.^14,15

PPSV vaccination is recommended in the United States for all persons aged ≥65 years and for adults aged <65 years at increased risk of invasive pneumococcal disease (cigarette smoking, chronic lung disease including asthma, chronic cardiovascular disease, diabetes mellitus, chronic liver disease, cirrhosis, chronic alcoholism, functional or anatomic asplenia, immunocompromising conditions including HIV infection, and diseases associated with treatment with immunosuppressive drugs or radiation therapy, cochlear implants and cerebrospinal fluid leaks).¹⁶ Similar recommendations are in place in most European Union countries, although in some countries vaccination is recommended only for adults with high risk conditions and not for healthy adults aged ≥65 years.¹⁷

A 13-valent conjugate pneumococcal vaccine (PCV13) has become available for adults, and the United States recommends that this new vaccine should be administered following by one dose of PPSV at least 8 weeks later in people aged ≥65 years.¹⁸ A recent randomized clinical trial among adults aged ≥ 65 y found that the PCV13 is effective against vaccine-type invasive and non-invasive community-acquired pneumonia in the elderly.¹⁹ In addition, the PCV has a clear effect against pneumococcal disease in the unvaccinated population (children and adults) that is not observed with the PPSV.²⁰

In Spain, free universal PPSV vaccination of people aged ≥65 years is recommended and is administered in primary care centers or hospitals throughout the year with no requirement for an order from the attending physician. However, most pneumococcal vaccination takes place during seasonal influenza vaccination campaigns.²¹ Although it was planned that all regions would introduce the pneumococcal conjugate vaccine (PCV) in pediatric vaccination calendars by the end of 2016, except for the Madrid Region during 2006-2012 and Galicia since 2011, no pneumococcal conjugate vaccine has been included in the pediatric vaccination calendar. However, it is recommended by the Spanish Association of Pediatrics and many private and public pediatricians recommend the vaccine, which is paid for the parents, in contrast with all routine vaccines included in the pediatric vaccination calendar, which are administered free of charge. The recommended schedule is 3 doses at 2, 4 and 6 months with a fourth dose between 12 and 18 months.²² The vaccine has been offered free of charge only for children with risk medical conditions for IPD. Pediatric PCV coverage, excluding Madrid and Galicia, is estimated at around 50%.²³

This coverage has been sufficient to originate some changes in the distribution of serotypes both in children and adults.²⁴ Studies carried out Catalan hospitals reported a progressive decline in rates of IPD in adults, mainly in PCV7 and PCV13 serotypes.^25,26 Guevara et al.²⁷ compared incidence rates of IPD before and after the introduction of the pediatric PCV13 in Navarra and found a decline of 81% in cases of IPD due to PCV13 serotypes in children aged <5 years and a 52% decline in the whole population.

The indirect protective effects (herd protection) of PCV in unvaccinated adults were reported in the United States a few years after the introduction of the vaccine.²⁸ Recent studies in countries with a high PCV coverage reinforce its role in avoiding cases of IPD caused by PCV serotypes in unvaccinated adults and the elderly.^29-36 Data on pneumococcal surveillance in the elderly must be assessed to decide the type of pneumococcal vaccine most appropriate to protect them against IPD. A population-based study by Ochoa et al.³⁷ in Tarragona found that the proportion of cases of IPD caused by PCV13 serotypes during 2006-2009 was 62.5% in people aged ≥ 65 y. Similar proportions (59.3% and 62.6%) were found in Spanish studies carried out in 2007-2009 ³⁸ and 2009.³⁹

The aim of this study was to investigate the factors associated with PPSV coverage in people aged ≥65 years hospitalized for causes unrelated to pneumonia, acute respiratory disease, or influenza-like illness in Spain.

Results

Of the 921 patients included in the study, 403 (43.8%) were vaccinated with pneumococcal vaccine: 394 had received the PPSV, 4 the PPSV and the PCV13 and 5 only the PCV13. Because the recommended schedule when conjugate vaccine is used is to administer PCV13 followed by PPSV at least 8 weeks later, the 5 patients who had received only the PCV13 were excluded from the analysis.

The distribution of the study variables (predisposing characteristics, enabling resources, and risk medical conditions) in vaccinated and unvaccinated patients is shown in Table 1. No differences were observed according to age and sex. Secondary or higher education level was more frequent in unvaccinated (61.1%) than in vaccinated patients (38.9%). Influenza vaccination in the 2013-14 season was more frequent in vaccinated (60.9%) than in unvaccinated patients (39.1%). A history of influenza vaccination in any of the 3 previous seasons was more frequent in vaccinated (59.4%) than in unvaccinated patients (40.6%). Patients who visited the general practitioner (GP) ≥3 times during the previous year had a higher vaccination rate than those who had not (51.1% and 28.8%, respectively). Patients making ≥3 hospital visits during the previous year had higher vaccination rates than those who did not (46.9% and 41.3%, respectively). Patients with a high level of dependence (Barthel index < 40) had a lower rate of vaccination than those with a low level of dependence (27.0% and 45.8%, respectively). Patients with low and high comorbidity had a higher vaccination rate than those without comorbidities (48.3%, 42.2%, and 40.4%, respectively). Table 2 shows the distribution of risk medical conditions in vaccinated and unvaccinated patients. Patients with chronic obstructive pulmonary disease (COPD) and chronic respiratory failure had a higher vaccination uptake (p < 0.01).

Table 1. Distribution of vaccinated and non-vaccinated hospitalized patients according to sociodemographic characteristics, comorbidities, and history vaccination between September 2013 and September 2014 from 7 Spanish regions.

	Vaccinated patients (N = 398)	Unvaccinated patients (N = 518)	Crude OR	p value
Predisposing characteristics
Age group
65-75	142 (44.9%)	174 (55.1%)	1
>75	256 (42.7%)	344 (57.3%)	1.05 (0.77 – 1.43)	0.77
Sex
Male	250 (44.9%)	307 (55.1%)	1
Female	148 (41.2%)	211 (58.8%)	0.84 (0.62 – 1.13)	0.25
Educational level
No education	183 (45.8%)	217 (54.2%)	1
Primary	122 (44.9%)	150 (55.1%)	0.86 (0.60 – 1.22)	0.40
Secondary or higher	93 (38.9%)	146 (61.1%)	0.66 (0.44 – 0.98)	0.04
Smoking status
Smoker or ex-smoker	202 (44.4%)	253 (55.6%)	1.04 (0.77 – 1.40)	0.79
Alcohol intake
Yes	10 (43.5%)	13 (56.5%)	0.68 (0.27 – 1.73)	0.42
Influenza vaccine in season 2013-14	240 (60.9%)	154 (39.1%)	5.52 (3.91 – 7.81)	<0.001
Influenza vaccine in season 2010-11	288 (61.8%)	178 (38.2%)	10.25 (6.98 – 15.06)	<0.001
Influenza vaccine in season 2011-12	297 (61.2%)	188 (38.8%)	10.12 (6.92 – 14.81)	<0.001
Influenza vaccine in season 2012-13	291 (61.3%)	184 (38.7%)	8.26 (5.75 – 11.87)	<0.001
Influenza vaccine in any of the 3 previous seasons	348 (59.4%)	238 (40.6%)	15.57 (10.29 – 23.54)	<0.001
Enabling resources
Marital status
Married/Cohabiting	235 (44.1%)	298 (55.9%)	1
Single	29 (43.9%)	37 (56.1%)	0.83 (0.46 – 1.51)	0.54
Widowed	126 (41.9%)	175 (58.1%)	0.97 (0.70 – 1.35)	0.87
Separated/Divorced	6 (60.0%)	4 (40.0%)	1.98 (0.47 – 8.26)	0.35
No. of GP visits
0-2	86 (28.8%)	213 (71.2%)	1
≥3	310 (51.1%)	297 (48.9%)	2.04 (1.46 – 2.85)	<0.001
No. of hospital visits
0-2	219 (41.3%)	311 (58.7%)	1
≥3	177 (46.9%)	200 (53.1%)	1.39 (1.02 – 1.89)	0.03
Household size
Live alone	70 (40.0%)	105 (60.0%)	1
Lives with cohabitant	328 (44.4%)	410 (55.6%)	1.16 (0.79 – 1.70)	0.45
Barthel index
≥40	367 (45.8%)	434 (54.2%)	1
<40	31 (27.0%)	84 (73.0%)	0.55 (0.34 – 0.89)	0.01
Charlson comorbidity index^*
0	69 (40.4%)	102 (59.6%)	1
1	115 (48.3%)	123 (51.7%)	1.81 (1.16 – 2.84)	0.01
≥2	214 (42.2%)	293 (57.8%)	1.58 (1.07 – 2.35)	0.02

* 0: no comorbidity, 1: low comorbidity, ≥2: high comorbidity

Table 2. Distribution of vaccinated and non-vaccinated hospitalized patients according to risk medical conditions between September 2013 and September 2014 from 7 Spanish regions.

	Vaccinated patients n (%), N = 398	Unvaccinated patients n (%), N = 518	Crude OR	p value
Chronic obstructive pulmonary disease	89 (57.8%)	65 (42.2%)	2.40 (1.59 – 3.62)	<0.001
Chronic respiratory failure	65 (52.0%)	60 (48.0%)	1.98 (1.25 – 3.14)	0.004
Asthma	48 (55.2%)	39 (44.8%)	1.15 (0.69 – 1.92)	0.59
Pneumonia during the last 2 years	22 (43.1%)	29 (56.9%)	1.08 (0.57 – 2.06)	0.80
Neoplasia	86 (43.7%)	111 (56.3%)	0.85 (0.60 – 1.22)	0.39
Transplantation	3 (60.0%)	2 (40.0%)	3.31 (0.47 – 23.49)	0.23
Immunosuppressive treatment	17 (37.0%)	29 (63.0%)	0.65 (0.32 – 1.29)	0.21
Asplenia	2 (66.7%)	1 (33.3%)	9.04 (0.53 – 153.65)	0.13
Diabetes	125 (39.4%)	192 (60.6%)	0.99 (0.72 – 1.36)	0.96
Renal failure	84 (42.9%)	112 (57.1%)	1.30 (0.90 – 1.88)	0.16
Nephrotic syndrome	4 (57.1%)	3 (42.9%)	1.51 (0.29 – 7.79)	0.62
Congestive heart disease	110 (42.8%)	147 (57.2%)	1.57 (1.09 – 2.25)	0.01
Disabling neurological disease	21 (32.8%)	43 (67.2%)	0.82 (0.45 – 1.51)	0.52
Obesity	96 (49.0%)	100 (51.0%)	1.38 (0.94 – 2.00)	0.10
Chronic liver disease	21 (50.0%)	21 (50.0%)	1.33 (0.65 – 2.72)	0.43
Hemoglobinopathy or anemia	61 (39.9%)	92 (60.1%)	1.00 (0.67 – 1.48)	0.98
Cognitive dysfunction	39 (35.8%)	70 (64.2%)	1.07 (0.66 – 1.72)	0.78
Convulsions	5 (38.5%)	8 (61.5%)	1.04 (0.28 – 3.83)	0.95
Neuromuscular disease	13 (52.0%)	12 (48.0%)	0.89 (0.38 – 2.11)	0.79

No vaccinated or unvaccinated patients presented AIDS or asymptomatic HIV infection.

The results of the multilevel regression model are shown in Table 3. Variables related to enabling resources (model 1) significantly associated with vaccine uptake were ≥3 GP visits during the previous year (OR=2.02; 95% CI 1.44-2.83) and a Barthel index < 40 (OR=0.48; 95% CI 0.29-0.78).

Table 3. Factors associated with 23-valent pneumococcal vaccination in hospitalized patients between September 2013 and September 2014 from 7 Spanish regions. Results of multilevel regression analysis.

	Model 1(Enabling resources)Adjusted OR (95%CI)	p value	Model 2(Model 1 + Predisposing characteristics)Adjusted OR (95%CI)	p value	Model 3(Model 2 + Risk medical conditions)Adjusted OR (95%CI)	p value
Enabling resources
No. of GP visits
0-2	1		1		1
≥3	2.02 (1.44 – 2.83)	<0.001	1.82 (1.27 – 2.62)	0.001	1.79 (1.25 – 2.57)	0.001
No. of hospital visits
0-2	1		1		1
≥3	1.37 (0.99 – 1.89)	0.05	1.42 (1.01 – 1.99)	0.04	1.37 (0.97 – 1.93)	0.07
Barthel index
≥40	1		1		1
<40	0.48 (0.29 – 0.78)	0.004	0.35 (0.19 – 0.63)	<0.001	0.33 (0.18 – 0.60)	<0.001
Predisposing characteristics
Influenza vaccine in season 2013/14			2.58 (1.73 – 3.85)	<0.001	2.57 (1.72 – 3.84)	<0.001
Influenza vaccine in any of the 3 previous seasons			11.60 (7.38 – 18.24)	<0.001	11.70 (7.42 – 18.45)	<0.001
Charlson comorbidity index
0					1
1					1.56 (0.96 – 2.55)	0.07
≥2					1.43 (0.92 – 2.24)	0.11

* 0: no comorbidity, 1: low comorbidity, ≥2: high comorbidity.

When predisposing characteristics were added (model 2), influenza vaccination in the 2013-14 season (OR=2.58; 95% CI 1.73-3.85) and influenza vaccination in any of the 3 previous seasons (OR=11.60; 95% CI 7.38-18.24) were associated with receiving the pneumococcal vaccination. Three or more GP visits and the Barthel index score were also associated with influenza vaccination.

When the Charlson index was added (model 3), the significant variables were the same as in model 2, with only slight decreases in the values of the odds ratio for ≥3 GP visits (OR=1.79; 95% CI 1.25-2.57); Barthel index < 40 (OR=0.33; 95% CI 0.18-0.60); influenza vaccination in the 2013-14 season (OR=2.57; 95% CI 1.72-3.84); and influenza vaccination in any of the 3 previous seasons (OR=11.70; 95% CI 7.42-18.45).

Discussion

This study found that pneumococcal vaccination coverage in the elderly in Spain is low (43.8%). Some Spanish studies have found a higher coverage. In Catalonia, where vaccination of the elderly was initiated at the end of 1999, rates slightly above 50% were reported in 2003 by Vila Córcoles et al.²¹ In the United States, achieving a 90% coverage of pneumococcal vaccination in non-institutionalized adults aged ≥65 years is a Healthy 2020 objective,⁴⁰ but the coverage is lower, even though vaccination is widely offered free to the elderly.⁴¹ In a nationally representative sample of non-institutionalized adults aged ≥65 years, PPSV coverage was 59.5% in 2013.⁴² In Ontario (Canada), the coverage in people aged ≥65 y in 2006 was 39% in healthy people and 49% in people with underlying diseases.⁴³ In the same country, Sabapathy et al.⁴⁴ found a PPSV coverage of 49.8% in 2009 and Schneeberg et al.⁴⁵ a coverage of 58% in a 2012 cross-sectional survey. In Israel, Schwartz et al.⁴⁶ reported a coverage of 72.7% during 2008-2009 in elderly members of a healthcare organization. In Australia, the coverage was 62.9% in 2006 in patients aged ≥ 60 y hospitalized in a large tertiary referral hospital,⁴⁷ and 67.6% in 2008 in people aged ≥65 years with chronic diseases (asthma, diabetes and cardiovascular disease).⁴⁸ The coverage found in patients admitted to a Korean university hospital in 2013 was 21.8%.⁴⁹

In patients aged ≥75 years admitted to a French geriatric unit during 2009-2010, PVC coverage was 17.2% but increased to 84.5% after an intervention that reminded physicians about whether pneumococcal vaccination was indicated or not.⁵⁰ In the same country, a coverage of 48% was found in 2013 in a large cohort of patients with secondary immune deficiency.⁵¹

ATurkish study by Arslam et al.⁵² found a PPSV coverage of 0.9% in elderly people in 2008 which increased to 19.1% after an intervention where families were asked whether their grandparents were vaccinated. In another Turkish study⁵³ carried out in 2009 in diabetic patients with a mean age of 57 y PPSV coverage was 9.8%, clearly lower that that found in the present study in diabetic patients (39.4%), but reached 40.7% after a physician training program.

In the present study, patients with COPD, asthma, and chronic respiratory failure had vaccination coverages of 57.8%, 55.2%, and 52.0%, respectively. Although not statistically significant, the coverage in patients with asthma was higher than in the whole population studied, but lower than that obtained by Dower et al.⁴⁸ Neither smokers nor patients with alcohol intake had a higher vaccination coverage.

In the bivariate analysis, we found an association between a higher Charlson index and PPSV uptake, but this disappeared in the final model.

Other variables that were associated with the PPSV coverage in the final model were having visited the GP ≥3 times during the last year, the Barthel index and having a history of influenza vaccination. Other authors^43,54 have also found that patients who had visited the physician more times during the last year are more likely to be vaccinated, but the study by Loubet et al.⁵¹ in immunocompromised patients did not find such an association. In our study, patients with a lower Barthel index (a higher level of dependence) had a lower rate of vaccination, probably because physicians do not believe that age alone is a good reason for recommending pneumococcal vaccination, independently of the functional status and limitations in activity. In a European survey of primary care physicians and specialists to determine pneumococcal disease awareness and attitudes, the patient's health condition was a key factor influencing a physician's decision to prescribe pneumococcal vaccination.⁵⁵

In contrast to the findings of Al-Sukhni et al.⁴³ a history of influenza vaccination in any of the 3 previous seasons was closely associated with pneumococcal vaccination in the final model (OR: 11.7; 95% CI 7.4-18.5), suggesting that, in Spain, patients who follow annual recommendations on influenza vaccination are more predisposed to accept other vaccine recommendations or that physicians who provide the influenza vaccination also provide pneumococcal vaccination. Liu et al.⁵⁶ in China and Loubet et al.⁵¹ in France found similar results to ours. We cannot say whether this may be due to the role of healthcare professionals or to patient attitudes as this was not an objective of the study. However, experience with the influenza vaccination⁵⁷ suggests that physicians' attitudes to pneumococcal vaccination might play an important role.

Studies^45,46,54 show that age is associated with vaccination, but we found no such association. Neither was gender associated with vaccination and the of other studies are heterogeneous: some found an association with male gender^44,46,58 and others with female gender.^45,54,59

Socioeconomic status is an important factor possibly associated with PPSV coverage in the elderly, but it is difficult to assess. We used the educational level as a proxy of socioeconomic status, and although higher coverages were found in patients with a lower educational level in the bivariate analysis, no association was shown in the final model. Scheenberg et al.⁴⁵ found lower coverages in people with a higher educational level and in people with higher incomes in the crude analysis but, in the adjusted analysis, the association disappeared. Sabapathy et al.⁴⁴ found a non-significant association between higher income and lower coverages in hospitalized elderly people. Other studies have found that a lower educational level⁵⁴ or lower income⁴⁶ are significantly associated with a low rate of vaccination in adjusted models.

The strategy and action plan for healthy aging in Europe states that a goal for vaccination of older people and infectious disease prevention is to reduce the health risks for older people that are due to gaps in vaccination against infectious diseases.⁶⁰ In order to reduce these gaps it is necessary to assess different possible strategies.

The results of a European physician survey found that risk medical conditions were a key factor in recommending the PPSV.⁵⁵ However, taking into account the expected increase in the population aged ≥65 years in forthcoming years and the frequency of risk medical conditions in age group, age-based policies should be potentiated. Some authors suggest that interventions such as physicians' independent initiation of standing orders, advertising, provider and patient mailing, reminder calls, easy access to patients' vaccination history, and patient and staff education might increase vaccination coverages.^50,61-63

The strategy followed in Spain of pneumococcal vaccination of the elderly without the need for an order from the attending physician, in order to vaccinate all persons aged ≥ 65 years, has been recommended by some authors^54,64 and we believe this strategy should not be changed in Spain. However, despite standing orders, physicians' opinions are a key factor in patients receiving the vaccine ^61,65 and staff education seems especially important in improving pneumococcal vaccination coverage in the elderly.

The role of physicians in promoting pneumococcal vaccination in the elderly has been widely recognized. A Spanish study by Picazo et al.⁶⁶ found that only 14% of people aged >60 years knew about the pneumococcal vaccine and 46% of unvaccinated elderly people stated the reason for not vaccinating was because the physician did not recommend it. In the survey by Lode et al.⁵⁵ the main driver for pneumococcal vaccination was recommendation from a healthcare professional. Therefore, the promotion of pneumococcal vaccination in primary care physicians and specialists by improving their knowledge of the burden of pneumococcal disease and their attitudes to pneumococcal vaccination might have a beneficial effect on vaccination coverages.

Strengths and limitations

Like any observational study, this work has strengths and limitations. The main strength of the study is that the vaccination status was obtained from written documents (hospital medical records, vaccination cards, or primary healthcare registers) and, therefore, it is unlikely that this information was biased. The differences found between the self-reported status and the true vaccination status might act as confounders of the main conclusions.^43,51 A possible limitation of the study is that it was made in hospitalized patients who are not representative of the total elderly population in Spain. However, these patients were non-institutionalized and were hospitalized for causes unrelated to pneumonia, acute respiratory disease, or influenza-like illness. Therefore, it may be suggested that there are no large differences with respect to the general elderly population. Another possible limitation is that no standard definitions of predisposing characteristics or enabling resources are available. We have followed criteria proposed by other authors^67,68 to examine whether the likelihood of vaccine uptake among the studied population is influenced by different factors. Finally, because this is a cross-sectional study, no causal relationship can be established. However, we identified some variables that are clearly associated with pneumococcal vaccination coverage and this may aid to improve vaccination strategies for the elderly.

Conclusion

The results of this study show that pneumococcal vaccination coverage of elderly people hospitalized for reasons other than pneumonia, acute respiratory disease or influenza-like illness is low in Spain and that some predisposing characteristics and enabling resources influence vaccination rates. The elderly should be a target for pneumococcal vaccination and healthcare workers should be encouraged to proactively propose vaccination. Educational campaigns aimed at the elderly could also help to increase the vaccination coverage and reduce the burden of pneumococcal disease in the community.

Methods

Study design

We carried out a cross-sectional study in hospitalized patients aged ≥65 years. 921 hospitalized patients from 19 hospitals located in the main cities of 7 Spanish regions (Andalusia, the Basque Country, Catalonia, Castile and Leon, Madrid, Navarre and Valencia Community) with unplanned hospital admission due to causes other than pneumonia, acute respiratory disease, or influenza-like illness were recruited between September 2013 and September 2014.

Patients included in the study were sought from patients admitted to the internal medicine service through the emergency department, and from patients admitted to the general surgery, otorhinolaryngology, ophthalmology, dermatology, and traumatology services. Patients referred from nursing homes and those who did not provide written consent were excluded.

Measures

Patients were considered vaccinated with the pneumococcal vaccine if they had received a dose of the vaccine at any time before data collection. Information on the vaccination status was obtained from vaccination registers, hospital medical records, vaccination cards or primary healthcare records.

Specifically-trained health professionals used a structured questionnaire to collect information by patient interview and review of medical records about predisposing characteristics, enabling resources and risk medical conditions. Social determinants of utilization are shown to affect the individual determinants both directly and through the health system. Various types of individual determinants then influence the health services used by the individual, determine influence the health services used by the individual.^67-69

The following predisposing characteristics were recorded: age, sex, educational level, smoking and alcohol intake, influenza vaccination status in the 2013-2014 season and influenza vaccination history in the 3 previous influenza seasons. Variables related to social support were collected to measure enabling resources: marital status, number of GP and hospital visits during the last year, whether the patient lived alone or at home with cohabitants, and the Barthel index, which has a total score ranging from 0 (complete dependence) to 100 (complete independence), as a measurement of limitations in activity in patients.⁷⁰ The Barthel index was used to assess the functional capacity at hospital admission. Risk medical conditions included were: COPD, chronic respiratory failure, history of pneumonia during the last 2 years, neoplasia, transplantation, immunosuppressive treatment, asplenia, diabetes, renal failure, nephrotic syndrome, autoimmune disease, AIDS, asymptomatic HIV infection, congestive heart disease, disabling neurological disease, obesity, chronic liver disease, hemoglobinopathy or anemia, cognitive dysfunction, convulsions and neuromuscular disease. Comorbidities were assessed using the Charlson comorbidity index, which assigns a weight to each comorbid condition.⁷¹

Statistical analyses

A bivariate analysis was made to compare vaccinated and unvaccinated patients taking into account the sociodemographic variables and risk medical conditions. As each Spanish region may introduce specific vaccination programs for specific population groups and because regions have some degree of autonomy in organizing health services, persons living in the same region tend to experience similar access to health care. Therefore, to estimate the crude and adjusted odds ratio (OR) we used multilevel regression models to consider the connection between the outcome variable in people from the same region, in order to obtain accurate statistical estimates of vaccine predictors. Covariates were introduced into the model using a backward stepwise procedure, with a cut-off point of p < 0.2.

Model 1 included only variables related to enabling resources; model 2 also included predisposing variables and model 3 included enabling resources, predisposing variables, and risk medical conditions.

The analysis was performed using the SPSS v.18 statistical package and the R v3.1.2 statistical software (http://cran.r-project.org).

Ethical considerations

All data collected were treated as confidential, in strict observance of legislation on observational studies. The study was approved by the Ethics Committees of the hospitals involved (Comité Ético de Investigación Clínica del Hospital Clínic de Barcelona; Comité Ético de Investigación Clínica del Hospital Universitari Mutua de Terrassa; Comité Ético de Investigación Clínica de la Corporació Sanitaria Parc Taulí de Sabadell; Comité Ético de Investigación Clínica del Hospital de Mataró, Consorci Sanitari del Maresme; Comité Ètic d'Investigació Clínica de la Fundació Unio Catalana Hospitals; Comité Ético de Investigación Clínica Área de Euskadi; Comité Ético de Investigación Clínica Área de Salud de Burgos y Soria; Comité Ético de Investigación Clínica Área de Salud de León; Comité Ético de Investigación Clínica Área de Salud Valladolid– Este; Comité Coordinador de Ética de la Investigación Biomédica de Andalucía; Comité Ético de Investigación Clínica del Hospital Ramón y Cajal, Madrid and Comité Ético de Investigación Clínica del Consorcio Hospital General Universitario de Valencia). Written informed consent was obtained from all the patients included in the study.

Abbreviations

CI	=	confidence interval
COPD	=	chronic obstructive pulmonary disease
GP	=	general practitioner
IPD	=	invasive pneumococcal disease
OR	=	odds ratio
PCV	=	pneumococcal conjugate vaccine
PCV7	=	7-valent conjugate pneumococcal vaccine
PCV13	=	13-valent conjugate pneumococcal vaccine
PPSV	=	23-valent pneumococcal polysaccharide vaccine

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

The other members of the Project PI12/02079 Working Group are:

J. Díaz-Borrego (Servicio Andaluz de Salud), A. Morillo (Hospital Universitario Virgen del Rocío), M.J. Pérez-Lozano (Hospital Universitario Virgen de Valme), J. Gutiérrez (Hospital Universitario Puerta del Mar), M. Pérez-Ruiz, M.A. Fernández-Sierra (Hospital Universitario San Cecilio y Virgen de las Nieves), S. Rojo-Rello, R. Ortiz de Lejarazu (Hospital Clínico Universitario de Valladolid), M.I. Fernández-Natal (Complejo Asistencial Universitario de León), T. Fernández-Villa (GIIGAS-Grupo de Investigación en Interacción Gen-Ambiente y Salud, Universidad de León), A. Pueyo (Hospital Universitario de Burgos), A. Vilella (Hospital Clínic), A. Antón (Hospital Universitari Vall d'Hebron; Universitat Autónoma de Barcelona), G. Navarro (Corporació Sanitària i Universitaria Parc Taulí), M. Riera (Hospital Universitari Mútua Terrassa), E. Espejo (Hospital de Terrassa), M.D. Mas, R. Pérez (ALTHAIA, Xarxa Hospitalaria de Manresa), J.A. Cayla, C. Rius (Agència de Salut Pública de Barcelona; CIBERESP), I. Crespo (CIBERESP, Universitat de Barcelona), C. Izquierdo, R. Torra (Agència de Salut Pública de Catalunya), M.F. Domínguez-Berjon, M.A. Gutiérrez, S. Jiménez, E. Gil, F. Martín, R. Génova-Maleras (Consejería de Sanidad), M.C. Prados, F. Ezzine de Blas (Hospital Universitario la Paz), J.C. Galan, E. Navas, L. Rodríguez (Hospital Ramón y Cajal), C.J. Álvarez, E. Banderas (Hospital Universitario 12 de Octubre), J. Chamorro (Complejo Hospitalario de Navarra), I. Casado, J. Díaz (Instituto de Salud Pública de Navarra), M.J. López de Goicoechea (Hospital de Galdakao), M. Morales (Universidad de Valencia; CIBERESP).

Funding

This work was supported by the National Plan of I+D+I 2008-2011 and ISCIII-Subdirección General de Evaluación y Fomento de la Investigación [Project PI12/02079] and cofunded by Fondo Europeo de Desarrollo Regional (FEDER. Unión Europea. Una manera de hacer Europa), and the Catalan Agency for the Management of Grants for University Research [AGAUR Grant number 2014/SGR 1403]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.