Incidence and Estimated Vaccine Effectiveness Against Hospitalizations for All-Cause Pneumonia Among Older US Adults Who Were Vaccinated and Not Vaccinated With 13-Valent Pneumococcal Conjugate Vaccine.

Importance: Following routine use of 13-valent pneumococcal conjugate vaccine (PCV13) in children in 2010, invasive pneumococcal disease rates have decreased substantially in children and adults. In 2014, the Advisory Committee for Immunization Practices recommended routine use of PCV13 among adults aged 65 years or older; previously only 23-valent pneumococcal polysaccharide vaccine (PPV23) was recommended. Objective: To estimate the association between the incidence of hospitalized all-cause pneumonia and lower respiratory tract infections (LRTI) and PCV13 vaccination among older adults at Kaiser Permanente Northern California (KPNC). Design, Setting, and Participants: This retrospective cohort study included adults at KPNC aged 65 years or older between July 1, 2015, and June 30, 2018, born after 1936 with no known history of PPV23 or PCV13 receipt before age 65. The study took place at an integrated health care system with an annual membership more than 4 million individuals, approximately 15% of whom are 65 years or older and broadly representative of the region. Data analysis took place from July 2018 to December 2021, and data collection took place from November 2016 to June 2018. Exposures: PCV13 vaccination status was ascertained from the electronic medical record (EMR). Individuals were considered vaccinated 14 days following immunization. Main Outcomes and Measures: First hospitalized all-cause pneumonia was identified in the EMR using primary/secondary discharge diagnosis International Classification of Diseases, Ninth Revision and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes. First hospitalized LRTI was identified using pneumonia codes and acute bronchitis codes. Relative risk (RR) of first pneumonia or LRTI hospitalization of individuals who were PCV13 vaccinated vs PCV13 unvaccinated was estimated using Poisson regressions adjusted for sex, race, ethnicity, age, influenza vaccine receipt, PPV23 receipt since age 65, pneumonia risk factors, health care use, and season. Vaccine effectiveness (VE) was estimated as (1-RR) × 100%.
Results: Of 192 061 adults, 107 957 (56%) were female and 139 024 (72%) were White individuals. PCV13 coverage increased from 0 in 2014 to 135 608 (76.9%) by 2018. There were 3488 individuals with 3766 pneumonia hospitalizations and 3846 individuals with 4173 LRTI hospitalizations. PCV13 was associated with an adjusted VE of 10.0% (95% CI, 2.4-17.0; P = .01) against hospitalized pneumonia and 9.4% (95% CI, 2.1-16.1; P = .01) against hospitalized LRTI. Conclusions and Relevance: In the context of a robust pediatric PCV13 immunization program, PCV13 vaccination of adults aged 65 years or older was associated with significant reductions in hospitalizations for all-cause pneumonia and LRTI. Vaccinating older adults with PCVs may provide broader public health benefit against pneumonia hospitalizations.

Introduction

Pneumonia can be caused by a wide range of bacteria, viruses, parasites, and fungi.[1] Streptococcus pneumoniae is the most common cause of bacterial pneumonia among adults and a leading cause of bacteremia and meningitis.[2] In the United States (US), more than 250 000 individuals are hospitalized annually for pneumonia, and approximately 50 000 of those hospitalizations result in death.[3] Adults aged 65 years or older are at higher risk of bacterial pneumonia, as well as individuals with immunocompromising conditions, diabetes, and chronic heart diseases.

Although there are 100 known S pneumoniae serotypes, relatively few cause most cases of pneumonia and invasive pneumococcal disease (IPD).[4,5] Before introduction of pneumococcal conjugate vaccines (PCVs), 90% of pediatric IPD in the US was because of the 13 serotypes included in PCV13.[6] The introduction of the 7-valent PCV to the routine infant vaccination schedule in the US in 2000, followed by PCV13 in 2010, has led to dramatic reductions in IPD in children, and substantial decreases in pneumococcal disease in adults via indirect protection.[7,8,9,10,11] Despite high PCV13 coverage among US children, the 13 serotypes in PCV13 continue to cause substantial burden of pneumonia among older US adults, particularly serotypes 3, 19A, and 19F.[12]

In 2012, the Advisory Committee on Immunization Practices (ACIP) recommended routine PCV13 in sequence with the 23-valent pneumococcal polysaccharide vaccine, (PPV23; originally licensed in 1983), to adults aged 19 years or older at high risk of IPD.[13] Based on a clinical trial in the Netherlands, Community Acquired Pneumonia Immunization Trial in Adults (CAPITA), that demonstrated 45.6% PCV13 efficacy against vaccine-type pneumococcal pneumonia in adults,[14] the ACIP expanded its recommendation in 2014 to administer PCV13 to all adults aged 65 years or older to address the remaining burden of disease.[2,11,15] However, evidence regarding the vaccine effectiveness (VE) of PCV13 against all-cause pneumonia in a real-world setting is limited.[16] A single US study, conducted by the US Centers for Disease Control and Prevention among US Medicare beneficiaries aged 65 years or older, reported that PCV13 effectiveness against community-acquired pneumonia (CAP) ranged from 6.0% to 11.4% depending on influenza season and influenza vaccine receipt.[17,18] Other studies have also shown PCV impact on lower respiratory tract infections (LRTI) beyond radiologically-confirmed CAP in adults[19,20] and children.[21,22,23,24] This study estimated the association between the incidence of hospitalized all-cause pneumonia and LRTI and use of PCV13 among adults aged 65 years or older within Kaiser Permanente Northern California (KPNC), a setting in which nearly 90% of children have been vaccinated with PCV13.[11]

Methods

The KPNC institutional review board (IRB) approved this cohort study and determined that informed consent could be waived under 45 CFR 46.116(f), as the required criteria were met. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.[26]

Setting

KPNC is an integrated health care delivery system with an annual membership of more than 4 million. Members receive nearly all their care at KPNC-owned facilities, which includes 259 medical clinics and 21 hospitals. KPNC’s electronic medical record (EMR) captures all inpatient, outpatient, and emergency department diagnoses, as well as all laboratory tests, therapeutic services, radiology tests, and medications. Vaccination data includes date, brand, lot, and anatomic site of injection. KPNC members comprise more than 30% of northern California’s population and are broadly representative of insured adults in the state with regard to underlying racial, ethnic, and socioeconomic demographics, although the very lowest incomes are underrepresented.[25] Fifteen percent of members are aged 65 years or older.

Study Population

This was a retrospective observational cohort study that included all KPNC members who were aged 65 years or older anytime between July 1, 2014, and June 30, 2018. The study end date was based on ACIP’s October 2018 change in recommendation for adults aged 65 years or older from routine PCV13 use to shared decision-making.[27] We defined each year of the study as beginning July 1 and ending June 30 of the following year. We required 12 months of continuous KPNC enrollment prior to each July 1 to capture prior health care use and comorbidities. We also required continuous enrollment between ages 58 to 64 years to capture receipt of PPV23 or PCV13 vaccination before age 65 years. Individuals could enter and exit the study population throughout the study period based on KPNC enrollment (eg, individuals who met enrollment criteria from July 1, 2016, to June 30, 2017, would be included for that year, but then subsequently excluded for 2017 to 2018 if their enrollment lapsed).

Before 2014, ACIP recommended pneumococcal vaccines for adults aged 19 years or older at high risk of IPD.[13] Exploratory analyses to assess whether adults vaccinated with PPV23 before age 65 years could confound the VE analyses revealed that following ACIP’s recommendation in 2014, adults who received PPV23 before age 65 years had both higher PCV13 coverage and higher rates of pneumococcal disease. We therefore only included people for whom we had full pneumococcal vaccine history and who received their first pneumococcal vaccine (PPV23 or PCV13) after age 65 years. Individuals who received PPV23 or PCV13 before age 65 years were excluded. For this reason, we also excluded those born before 1936 because of missing prior vaccination history in the EMR, including PPV23 vaccine status.

Outcomes

Study outcomes were hospitalized all-cause pneumonia and LRTI. We defined all-cause pneumonia using primary and secondary hospital discharge diagnosis International Classification of Diseases, Ninth Revision (ICD-9) codes 480–486, 487.0, and 519.8, or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes J10.0, J11.0, J12–J18 (except J18.2), and J22. We defined LRTI using the above codes for all-cause pneumonia, in addition to primary or secondary discharge diagnosis codes for acute bronchitis (ICD-9 codes 466.0, 490, and 491.22, or ICD-10 codes J20, J40, and J44.0).

Covariates

Demographic covariates included sex, race, ethnicity, and age (as of each season starting July 1), as these covariates are associated with an increased risk of pneumococcal disease.[28,56] Race (ie, Asian, Black, Hawaiian or Pacific Islander, Native American or Alaskan Native, and White) and ethnicity (ie, Hispanic and non-Hispanic) were based on self-report as identified in the EMR. Clinical covariates included the number of weeks with an outpatient visit in the prior year as an indicator of health care use (low: 0-1 week; medium: 2-3 weeks; high: ≥4 weeks), Charlson comorbidity index (low: 0; medium: 1; high: ≥2),[29] PPV23 vaccine receipt since age 65 years (within or beyond 5 years), whether each month was during the influenza season (October through April),[30,31] and receipt of influenza vaccine in the prior and current season.[17] We also included covariates associated with increased risk of pneumococcal disease, including history of asthma, chronic obstructive pulmonary disease (COPD), diabetes, coronary heart disease, and pneumonia (eTable 1 in the Supplement).

Statistical Analysis

We calculated yearly incidence rates per 100 000 person-years of the first all-cause pneumonia or LRTI hospitalization event for each July to June season. Poisson confidence intervals were computed for these incidence rates.

We estimated the relative risk (RR) of all-cause pneumonia and LRTI by comparing the risk among individuals who were vaccinated to the risk among individuals who were unvaccinated using multivariable adjusted Poisson regression. Vaccination status with PCV13 was an independent variable in this model. We ascertained PCV13 vaccination status and pneumonia and LRTI hospitalizations on a monthly basis and categorized individuals as vaccinated 14 days following immunization. We summarized all person-years, vaccine exposure, and outcome incidence by calendar month. When there was an outcome event and a vaccination in the same month, we accounted for the order of occurrence. We calculated VE as (1–RR) × 100%. All P values were 2-sided, and significance was defined as P < .05.

We limited analyses to July 1, 2015, through June 30, 2018, because PCV13 uptake was less than 10% from July 1, 2014, through June 30, 2015. To ensure that results were robust to the study period of 2015 to 2018, we conducted sensitivity analyses which extended the study period to 2014 to 2018.

Only the first hospitalized pneumonia or LRTI event during the study period contributed to the primary analyses because pneumonia hospitalization itself could be the catalyst for receipt of PCV13, potentially biasing assessment of VE. To allow individuals to be included in more than 1 season, we conducted secondary analyses in which individuals could contribute the first hospitalized pneumonia or LRTI event per yearly season (July to June). Secondary analyses used robust standard error computation to calculate CIs that accounted for repeated events in the same individual.

All data analyses were performed using SAS software version 9.4 (SAS Institute, Inc). Data analysis took place from July 2018 to December 2021, and data collection took place from November 2016 to June 2018.

Results

Between July 1, 2015, to June 30, 2018, the study population included 192 061 individuals (Figure), 109 957 (56%) of whom were female and 26 233 (14%) were Asian individuals, 12 741 (7%) were Black individuals, and 139 024 (72%) were White individuals (Table 1). Most demographic characteristics were stable across the study period (Table 2). The median age was 69 years (minimum to maximum age, 65-78 years) at the start of the study (July 1, 2015), and 71 years (minimum to maximum age, 65-81 years) by the end of the study (June 30, 2018). Most people included in the analysis were aged 65 to 69 years in 2015 and aged 70 to 79 years by 2018, with only 5% aged 80 years or older by 2018 (because of the 1936 birth year restriction).

Figure.

Kaiser Permanente Northern California (KPNC) Study Population

PCV13 indicates 13-valent pneumococcal conjugate vaccine; PPV23, 23-valent pneumococcal polysaccharide vaccine.

Table 1.

Person-years at Risk by Sociodemographic and Health Care Use Characteristics in Adults Aged 65 Years or Older at KPNC, July 1, 2015, to June 30, 2018

Characteristics	Person-years, No.				No. (%)a (N=192 061)
	All-cause pneumonia		LRTI
	PCV13 vaccinated	Not PCV13 vaccinated	PCV13 vaccinated	Not PCV13 vaccinated
Sex
Male	87 906	129 117	87 831	128 929	84 104 (43.8)
Female	113 788	167 600	113 712	167 403	107 957 (56.2)
Race
Asian	23 429	43 593	23 417	43 549	26 233 (13.7)
Black	16 885	15 977	16 868	15 947	12 741 (6.6)
Hawaiian/Pacific Islander	853	1165	852	1160	801 (0.4)
Native American/Alaskan	1107	1384	1104	1383	967 (0.5)
Unknown	14 888	15 628	14 887	15 612	12 295 (6.4)
White	144 531	218 970	144 415	218 680	139 024 (72.4)
Ethnicity
Hispanic	18 614	26 099	18 601	26 068	17 530 (9.1)
Non-Hispanic	148 069	241 300	147 956	240 986	147 972 (77.0)
Unknown	35 011	29 319	34 987	29 278	26 559 (13.8)
Age group
65-69 years	110 918	128 770	110 872	128 692	NA
70-79 years	90 271	165 632	90 168	165 332	NA
≥80 years	505	2315	502	2308	NA
Received PPV23 (since age 65)
Yes	104 047	215 253	103 939	214 924	NA
No	97 647	81 464	97 604	81 408	NA
Prior health care use
Low	113 851	154 875	113 791	154 702	NA
Medium	57 328	95 585	57 285	95 473	NA
High	30 515	46 257	30 468	46 157	NA
Charlson score
Low	130 502	156 980	130 477	156 945	NA
Medium	42 370	80 586	42 332	80 518	NA
High	28 822	59 151	28 734	58 870	NA
Received flu vaccine, current season	73 761	200 909	73 700	200 647	NA
Received flu vaccine, prior seasonb	97 212	237 987	97 134	237 685	NA
History of asthma	33 165	58 347	33 102	58 124	NA
History of coronary heart disease	21 728	38 559	21 666	38 379	NA
History of COPD	7608	13 318	7556	13 138	NA
History of diabetes	27 266	43 094	27 204	42 951	NA
History of pneumonia	21 039	37 296	20 963	37 132	NA
Flu season (October-April)	112 011	178 568	111 922	178 340	NA
Not flu season (May-September)	89 683	118 150	89 621	117 992	NA
July 1, 2015, through June 30, 2016	97 692	56 329	97 638	56 298	NA
July 1, 2016, through June 30, 2017	56 910	109 634	56 858	109 511	NA
July 1, 2017, through June 30, 2018	47 092	130 754	47 046	130 523	NA

Abbreviations: COPD, chronic obstructive pulmonary disease; KPNC, Kaiser Permanente Northern California; LRTI, lower respiratory tract infections; PCV13, 13-valent pneumococcal conjugate vaccine; PPV23, 23-valent pneumococcal polysaccharide vaccine.

Unique Ns are only included for static variables that are not time-varying to avoid duplicate counting.

Prior flu season is based on July through June of prior yearly cycle.

Table 2.

Demographic Characteristics of Study Population, Adults ≥65 Years Old at KPNC, July 1, 2014, to June 30, 2018, by PCV Vaccination

Characteristics	July 1, 2014, (N = 142 020), No. (%)		July 1, 2015, (N = 155 027), No. (%)		July 1, 2016, (N = 167 676), No. (%)		July 1, 2017, (N = 179 145), No. (%)		June 30, 2018, (N = 176 388), No. (%)
	PCV13	No PCV13	PCV13	No PCV13	PCV13	No PCV13	PCV13	No PCV13	PCV13	No PCV13
PCV13 vaccinated, No. (%) of total	0	142 020 (100)	15 816 (10.2)	139 211 (89.8)	92 297 (55.0)	75 379 (45.0)	123 210 (68.8)	55 935 (31.2)	135 608 (76.9)	40 780 (23.1)
PPV23 vaccinated	0	101 269 (71.3)	10 794 (68.2)	97 374 (69.9)	73 026 (79.1)	36 734 (48.7)	96 998 (78.7)	18 088 (32.3)	111 547 (82.3)	12 450 (30.5)
Sex
Male
Female	0	79 467 (56.0)	8898 (56.3)	78 299 (56.2)	52 150 (56.5)	42 402 (56.3)	69 636 (56.5)	31 680 (56.6)	76 754 (56.6)	23 242 (57.0)
Race
Asian	0	17 835 (12.6)	2581 (16.3)	17 614 (12.7)	13 557 (14.7)	8920 (11.8)	18 176 (14.8)	6484 (11.6)	20 117 (14.8)	4283 (10.5)
Black	0	9466 (6.7)	635 (4.0)	9638 (6.9)	4806 (5.2)	6301 (8.4)	6798 (5.5)	4963 (8.9)	7594 (5.6)	3933 (9.6)
Hawaiian/Pacific Islander	0	545 (0.4)	58 (0.4)	550 (0.4)	350 (0.4)	329 (0.4)	500 (0.4)	246 (0.4)	569 (0.4)	166 (0.4)
Native American/Alaskan	0	694 (0.5)	63 (0.4)	720 (0.5)	425 (0.5)	414 (0.5)	580 (0.5)	305 (0.5)	651 (0.5)	226 (0.6)
Unknown	0	7852 (5.5)	659 (4.2)	8327 (6.0)	4681 (5.1)	5516 (7.3)	6680 (5.4)	4742 (8.5)	7659 (5.6)	3674 (9.0)
White	0	105 628 (74.4)	11 820 (74.7)	102 362 (73.5)	68 478 (74.2)	53 899 (71.5)	90 476 (73.4)	39 195 (70.1)	99 018 (73.0)	28 498 (69.9)
Ethnicity
Hispanic	0	12 409 (8.7)	1137 (7.2)	12 505 (9.0)	8028 (8.7)	6983 (9.3)	11 016 (8.9)	5295 (9.5)	12 336 (9.1)	3753 (9.2)
Non-Hispanic	0	113 504 (79.9)	13 284 (84.0)	109 324 (78.5)	75 851 (82.2)	55 275 (73.3)	99 640 (80.9)	38 752 (69.3)	108 498 (80.0)	27 700 (67.9)
Unknown	0	16 107 (11.3)	1395 (8.8)	17 382 (12.5)	8418 (9.1)	13 121 (17.4)	12 554 (10.2)	11 888 (21.3)	14 774 (10.9)	9327 (22.9)
Age group, y
65-69	0	75 635 (53.3)	7870 (49.8)	71 338 (51.2)	39 625 (42.9)	42 102 (55.9)	47 321 (38.4)	32 359 (57.9)	43 870 (32.4)	19 538 (47.9)
70-79	0	66 385 (46.7)	7946 (50.2)	67 873 (48.8)	52 672 (57.1)	33 277 (44.1)	73 623 (59.8)	22 966 (41.1)	84 397 (62.2)	19 821 (48.6)
≥80	NA	NA	NA	NA	NA	NA	2266 (1.8)	610 (1.1)	7341 (5.4)	1421 (3.5)
Prior health care use
Low	0	65 523 (46.1)	7310 (46.2)	72 844 (52.3)	45 308 (49.1)	44 564 (59.1)	65 849 (53.4)	34 443 (61.6)	73 800 (54.4)	25 123 (61.6)
Medium	0	47 500 (33.4)	5681 (35.9)	43 106 (31.0)	31 437 (34.1)	20 320 (27.0)	38 947 (31.6)	14 311 (25.6)	42 149 (31.1)	10 411 (25.5)
High	0	28 997 (20.4)	2825 (17.9)	23 261 (16.7)	15 552 (16.8)	10 495 (13.9)	18 414 (14.9)	7181 (12.8)	19 659 (14.5)	5246 (12.9)
Charlson score
Low	0	92 849 (65.4)	9948 (62.9)	91 560 (65.8)	45 923 (49.8)	47 683 (63.3)	57 904 (47.0)	35 303 (63.1)	66 940 (49.4)	25 677 (63.0)
Medium	0	31 468 (22.2)	3944 (24.9)	30 389 (21.8)	26 378 (28.6)	15 591 (20.7)	35 718 (29.0)	11 526 (20.6)	38 319 (28.3)	8484 (20.8)
High	0	17 703 (12.5)	1924 (12.2)	17 262 (12.4)	19 996 (21.7)	12 105 (16.1)	29 588 (24.0)	9106 (16.3)	30 349 (22.4)	6619 (16.2)
Received flu vaccine, prior seasonb	0	99 002 (69.7)	12 820 (81.1)	91 795 (65.9)	77 128 (83.6)	36 224 (48.1)	98 626 (80.0)	20 793 (37.2)	105 518 (77.8)	12 185 (29.9)
Comorbidities
Asthma	0	24 871 (17.5)	2859 (18.1)	25 023 (18.0)	18 350 (19.9)	12 889 (17.1)	24 072 (19.5)	7399 (13.2)	26 283 (19.4)	5609 (13.8)
Coronary heart disease	0	14 017 (9.9)	1625 (10.3)	13 929 (10.0)	12 085 (13.1)	8435 (11.2)	16 113 (13.1)	5291 (9.5)	17 738 (13.1)	4141 (10.2)
COPD	0	5391 (3.8)	554 (3.5)	5119 (3.7)	4046 (4.4)	2813 (3.7)	5461 (4.4)	1827 (3.3)	6331 (4.7)	1473 (3.6)
Diabetes	0	18 625 (13.1)	1934 (12.2)	18 981 (13.6)	13 114 (14.2)	10 052 (13.3)	18 018 (14.6)	6468 (11.6)	20 062 (14.8)	5117 (12.5)
Pneumonia	0	15 658 (11.0)	1896 (12.0)	15 454 (11.1)	11 335 (12.3)	7658 (10.2)	15 519 (12.6)	5528 (9.9)	17 731 (13.1)	4145 (10.2)

Abbreviations: COPD, chronic obstructive pulmonary disease; KPNC, Kaiser Permanente Northern California; PCV13, 13-valent pneumococcal conjugate vaccine; PPV23, 23-valent pneumococcal polysaccharide vaccine.

Population totals vary depending on the date selected, as members may enter and exit the cohort each annual season.

Prior flu season is based on July through June of prior yearly cycle.

The proportion of individuals with comorbidities in individuals who were vaccinated vs unvaccinated was comparable at the beginning of the study period (eg, 1625 [10.3%] of individuals with coronary heart disease who were vaccinated vs 13 929 [10.0%] of individuals with coronary heart disease who were unvaccinated on July 1, 2015). Over time, the proportion with comorbidities was slightly higher in PCV13 individuals who were vaccinated, and especially in those with asthma (ie, 26 283 [19.4%] of individuals who were vaccinated vs 5609 [13.8%] of individuals who were unvaccinated had asthma). Throughout the study period, the proportion of individuals who were vaccinated and received an influenza vaccine in the prior year was higher (77.8% to 83.6% depending on the year) than among individuals who were unvaccinated (29.9% to 65.9% depending on the year).

From 2015 to 2018, PCV13 vaccine coverage increased from 0 in 2014 to 135 608 of 176 388 individuals (76.9%) of the study population (Table 1). PPV23 vaccine coverage remained stable (eg, 71.3% on July 1, 2014; 69.8% on July 2, 2015; 65.5% on July 1, 2016; 64.2% on July 1, 2017; and 70.3% on June 30, 2018).

Pneumonia and LRTI Incidence

Allowing 1 hospitalization event across 2015 to 2018, there were 3488 pneumonia hospitalizations and 3846 LRTI hospitalizations. Allowing 1 hospitalization annually per July to June season, there were 3488 unique individuals with 3766 pneumonia hospitalizations and 3846 unique individuals with 4173 LRTI hospitalizations. Repeat hospitalizations (ie, up to 1 hospitalization annually over 3 July to June seasons, for a maximum of 3 hospitalizations) for pneumonia or LRTI were relatively uncommon (pneumonia: 236 individuals with 2 hospitalizations and 21 individuals with 3 hospitalizations; LRTI: 273 individuals with 2 hospitalizations and 27 individuals for 3 hospitalizations).

Across the study period, the incidence rate (IR) of first event of hospitalized pneumonia per yearly season for individuals aged 65 years or older ranged from 704.3 to 809.4 per 100 000 person-years and increased with older age (eTable 2 in the Supplement). Among individuals excluded from the final study population, hospitalized pneumonia rates among adults born before 1936 ranged from 3757.8 to 4643.9 per 100 000 person-years (eTable 3 in the Supplement), and among high risk adults born after 1936 and who received PPV23 before age 65, rates ranged from 2107.4 to 2589.0 per 100 000 person-years (eTable 4 in the Supplement). Throughout the study period, unadjusted IRs of first event hospitalized pneumonia varied substantially by year with no consistent trend by vaccination status and age group. IRs were higher for hospitalized LRTI than pneumonia, and overall trends by age and year were similar to pneumonia.

VE Analyses

The primary analysis included 192 061 unique individuals (Figure). PCV13 was associated with adjusted VE of 10.0% (95% CI, 2.4%-17.0%; P = .01) against pneumonia and 9.4% (95% CI, 2.1%-16.1%; P = .01) against LRTI (Table 2). Sex, age, health care use, having comorbidities, receipt of influenza vaccine in the prior season, and receipt of PPV23 more than 5 years prior were significantly associated with pneumonia or LRTI hospitalization (eTable 5 in the Supplement and eTable 6 in the Supplement). Secondary analyses, which allowed each individual to contribute only the first hospitalization annually, were similar, with adjusted VE of 9.6% (95% CI, 2.1%-16.6%; P = .01) against pneumonia and 8.7% (95% CI, 1.5%-15.4%; P = .02) against LRTI (Table 3). Sensitivity analyses which included all years 2014 to 2018 for both the primary and secondary analyses minimally impacted VE estimates (eTable 4 in the Supplement).

Table 3.

Association Between PCV13 Vaccination and Hospitalized Pneumonia and Lower Respiratory Tract Infections in Adults ≥65 Years Old at KPNC, July 1, 2015 to June 30, 2018

PCV13 vaccinated vs unvaccinated	Unadjusted RR (95% CI)	Adjusted RR (95% CI)a,b	Estimated adjusted vaccine effectiveness % (95% CI)	P value
Primary analyses (first event only)
Pneumonia hospitalization	1.07 (1.003-1.15)	0.90 (0.83-0.98)	10.0 (2.4-17.0)	.01
LRTI hospitalization	1.07 (1.004-1.14)	0.90 (0.84-0.98)	9.4 (2.1-16.1)	.01
Secondary analyses (first event each yearly season)
Pneumonia hospitalization	1.097 (1.02-1.18)	0.90 (0.83-0.98)	9.6 (2.1-16.6)	.01
LRTI hospitalization	1.15 (1.08-1.22)	0.91 (0.85-0.99)	8.7 (1.5-15.4)	.02

Abbreviations: KPNC, Kaiser Permanente Northern California; LRTI, lower respiratory tract infections; PCV13, 13-valent pneumococcal conjugate vaccine; PPV23, 23-valent pneumococcal polysaccharide vaccine; RR, relative risk.

All models have been adjusted for PPV23 vaccination status, sex, race, ethnicity, age, prior health care use, Charlson score, influenza vaccine receipt in the prior season, influenza vaccine receipt in the current season, influenza season, history of asthma, history of chronic obstructive pulmonary disease, history of diabetes, history of coronary heart disease, and history of pneumonia.

The following covariates were statistically significant (α < 0.05) in all adjusted models: PPV23 vaccination more than 5 years prior, sex, age, high values for prior health care use and Charlson score, influenza vaccine receipt in the prior season, flu season (Oct-April), and history of COPD, diabetes, coronary heart disease, and pneumonia; for the LRTI hospitalization outcome, in addition to these covariates, having a history of asthma was also statistically significant. Please refer to eTable 5 and eTable 6 in the Supplement for additional detail.

Discussion

This study evaluated a population of adults in a setting where PCV13 had been routinely administered to infants since 2010 and PPV23 to adults since the 1980s. We found that vaccination with PCV13 was associated with a 10.0% reduction against hospitalized pneumonia and 9.4% reduction against hospitalized LRTI when compared with adults who were unvaccinated. This study provides evidence that direct PCV13 vaccination of older adults may be associated with broad public health benefit in preventing hospitalizations associated with all-cause pneumonia and LRTI, even in a setting with long-standing PCV13 vaccination of children with very high coverage.[11]

These results are consistent with other studies. The estimated VE of 10% against all-cause hospitalized pneumonia is similar to the 8% found in a post hoc analysis of the CAPITA randomized control trial in the Netherlands.[32] The KPNC population is comparable to the Netherlands in that both have very high pediatric PCV13 vaccine coverage, with Netherlands having approximately 96% 10-valent PCV vaccine coverage[32] and KPNC having nearly 90% PCV13 coverage (compared with 82% nationally).[33] Current results are also consistent with an observational study conducted by the US Centers for Disease Control and Prevention, which analyzed US Medicare beneficiary data and estimated a PCV13 VE of 6% to 11% against all-cause CAP hospitalization.[15,17] Their rates of CAP followed a similar trend as our analysis with increasing incidence by age.[15,17] One study from Italy also reported a 20% relative reduction in the incidence of emergency department LRTI events 1 year following the introduction of PCV13 vaccine in adults aged 70 to 75 years.[34] However, this analysis excluded individuals who received PPV23 in the year prior to PCV13 vaccination and PPV23 vaccine coverage may not be comparable to the current study.[35] A population-based cohort study in Germany also reported a 12% relative risk reduction in the incidence of all-cause inpatient and outpatient pneumonia after PCV13 vaccination following its routine use in adults in 2012, though PCV13 coverage was substantially lower (approximately 2%) than the current study.[36] Finally, controlling for PCV13 vaccination status, the current findings are consistent with evidence that PPV23 vaccination may not be effective against all-cause pneumonia.[37,38,39,40,41,42,43,44,45,46,47]

In June 2019, the ACIP concluded that there have been no additional population-level reductions in PCV13-type pneumonia that would support continued PCV13 vaccination in older adults based on evaluation of overall trends in pneumonia incidence pre- and post-PCV13 introduction in adults.[27] Our observations regarding trends in pneumonia incidence are consistent with these reports. However, hospitalized pneumonia incidence over time is a summary measure combining direct and indirect PCV13 effects, vaccine coverage, and the impact of other respiratory pathogens, particularly yearly variations in influenza (and influenza VE) and respiratory syncytial virus; such analyses cannot measure vaccine effectiveness among individuals who were vaccinated. After adjustment for factors that can confound the association between PCV13 vaccination and risk of IPD,[17,28,30,31,48] our results indicate that PCV13 vaccination of adults was associated with reductions in all-cause hospitalized pneumonia and LRTI in persons aged 65 years and older.

In October 2021, the ACIP recommended routine use of higher valent PCVs in all adults aged 65 years or older (either PCV20 alone or PCV15 followed by PPV23).[49,50] While new PCVs will likely further reduce the burden of vaccine-type IPD, no studies have directly compared the effectiveness of PCV15 vs PCV20. Given the current finding that PCV13 was associated with lower incidence of pneumonia and LRTI hospitalizations, future studies assessing whether higher valent PCVs reduce IPD caused by vaccine-type serotypes as well as clinically meaningful outcomes, such as all-cause pneumonia and LRTI, will be important.

Based on both the CAPITA trial, which found a 46% efficacy against vaccine-type CAP, and an observational study among older US adults, which detected PCV13 serotypes in 4% of adults aged 65 years or older diagnosed with radiologically-confirmed CAP,[14,51] one might have expected a 2% reduction in all-cause pneumonia hospitalizations rather than the 10% reduction observed. This is likely to be at least in part because of the use of ICD-10 codes to identify pneumonia without requiring radiologic confirmation in the current study. Another reason may be that the serotype-specific urinary antigen detection test used to measure CAP serotype distribution in the earlier studies may be insensitive against nonbacteremic pneumonia.[19] Alternatively, relatively new data suggest PCV13 was associated with a reduction in virus-associated respiratory disease outcomes in children younger than 2 years and adults,[19,52,53] possibly because of interactions between viruses and pneumococci in the upper airway. The finding that prior season influenza vaccination was significantly associated with a reduced likelihood of all-cause pneumonia and LRTI may support this hypothesis, although further investigation into this potential association would be required. Unmeasured confounding also cannot be ruled out. A more definitive explanation for this discrepancy is urgently needed.

Strengths of the current study include the large size of our population and the large number of hospitalizations in the VE analyses. In addition, KPNC has very high pediatric coverage with PCV13, allowing us to assess additional PCV13 effects in adults in the context of high pediatric coverage. Furthermore, PCV13 vaccination rates in our study population of older adults were nearly 77% by the end of the study period, which surpassed the national coverage estimate of 47% in 2018.[18] In addition and in contrast to other studies, we were able to finely categorize PPV23 vaccination status before age 65 years with a high degree of confidence to account for confounding related to being at higher risk of pneumococcal disease. Finally, prior studies of PCV13 effectiveness focused on vaccine serotype-specific laboratory outcomes.[14,54] In contrast, the current study focused on all-cause pneumonia and LRTI hospitalizations and found that PCV13 was associated with decreased incidence for both clinically important outcomes, which suggests that use of PCV13 in adults may benefit patients and public health.

Limitations

This study had limitations. While we accounted for individuals who had been vaccinated with PPV23 before and since the ACIP recommendation, we were unable to fully assess the effectiveness of PCV13 in isolation because the analyses were unable to account for different possible vaccine sequences (eg, PCV13-PPV23 vs PPV23-PCV13 vs PCV13 and multiple PPV23 doses). Also, there may have been residual confounding. For example, alcoholism and cigarette smoking, which are associated with pneumococcal disease, were not adjusted for because of EMR data limitations. There was likely still residual confounding related to differences in health care seeking behavior and in baseline health among individuals born after 1936 who were not vaccinated prior to age 65 and between those who received PCV13 earlier vs later in the study period. Finally, this study did not include older KPNC members (born before 1936) for whom pneumococcal vaccine may have had a greater benefit, as older age was associated with an increased risk of pneumococcal disease.

Conclusions

This study’s findings suggest that in the setting of a robust pediatric PCV13 immunization program, direct PCV13 vaccination of adults aged 65 years or older was associated with substantial additional protection against all-cause pneumonia and LRTI hospitalizations. Preventing hospitalizations associated with pneumonia by increasing the proportion of adults who were vaccinated with PCVs may have broader public health benefit, such as reductions in chronic disease exacerbations. As recommendations regarding pneumococcal vaccination of adults aged 65 years or older continue to evolve, especially with higher valent PCVs recently approved,[55] it will be important to monitor the impact of vaccination policies on both etiologically confirmed vaccine-serotype pneumococcal disease and all-cause pneumonia. Consideration of vaccine effectiveness against all-cause pneumonia hospitalizations could provide additional perspectives when evaluating the public health benefits of adult and pediatric PCV programs.