Cost-effectiveness of implementing 13-valent pneumococcal conjugate vaccine for U.S. adults aged 19 years and older with underlying conditions

ABSTRACT

In June 2019, the Advisory Committee on Immunization Practices (ACIP) changed the recommendation for routine 13-valent pneumococcal conjugate vaccine (PCV13) use in immunocompetent adults aged ≥65 years, including those with select chronic medical conditions (CMC). ACIP now recommends PCV13 for this group of adults based on shared clinical decision-making. Because adults with CMC continue to be at increased risk for pneumococcal disease, we assessed the cost-effectiveness of administering PCV13 in series with the recommended 23-valent pneumococcal polysaccharide vaccine (PPSV23) for adults aged ≥19 years with CMC.

We used a probabilistic model following a cohort of 19-year-old adults. We used Monte Carlo simulation to estimate the impact on program, medical, and non-medical costs (in 2017 U.S. dollars [$], societal perspective), and pneumococcal disease burden when administering PCV13 in series with PPSV23. We used PCV13 efficacy and post-licensure vaccine effectiveness (VE) data to estimate VE against PCV13 type disease (separately for disease by serotype 3 [ST3], the most common PCV13 type, and all other PCV13 serotypes). We considered a range of estimates for sensitivity analyses. Analyses were performed in 2019.

In the base case, assuming no PCV13 effectiveness against ST3 disease, adding a dose of PCV13 upon CMC diagnosis cost $689,299 per QALY gained. This declined to $79,416 per QALY if VE against ST3 was estimated to be equivalent to other PCV13-types.

Administering PCV13 in series with the recommended PPSV23 for adults with CMC was not cost saving. Results were sensitive to estimated PCV13 VE against ST3 disease.

KEYWORDS:

• 13-valent pneumococcal conjugate vaccine
• united States
• adults
• chronic underlying conditions
• risk-based recommendation

Introduction

In the United States, approximately 30,000 cases of invasive pneumococcal disease (IPD) such as pneumococcal meningitis and bacteremia occur annually.¹ People who are at the extremes of age (<2 years, ≥65 years)² and those with select underlying medical conditions are at increased risk of pneumococcal disease.³ Introduction of pneumococcal conjugate vaccines into the routine infant immunization program in the United States led to substantial decline in IPD caused by vaccine serotypes, not only among vaccinated children, but also in unvaccinated older individuals through indirect effects.^4,5 Since 2012, U.S. adults aged ≥19 years with immunocompromising conditions, cerebrospinal fluid leaks, or cochlear implants have been recommended to receive 13-valent pneumococcal conjugate vaccine (PCV13) in series with 23-valent polysaccharide vaccine (PPSV23).⁶ A previous cost-effectiveness analysis showed that such intervention would be cost saving.⁷ Adults aged ≥19 years with chronic medical conditions (CMC), such as chronic heart, lung, liver disease, diabetes mellitus, alcoholism, and cigarette smoking, are recommended to receive PPSV23 only. In 2014, PCV13 was recommended in series with PPSV23 for adults aged ≥65 years (including those with CMC).⁸ In June 2019, the Advisory Committee on Immunization Practices (ACIP) voted to remove the recommendation for routine PCV13 use in immunocompetent adults aged ≥65 years in light of additional years of data showing the impact of herd effects from childhood immunization; a cost-effectiveness analysis showed that continuing the recommendation in a setting of observed to date indirect effects from PCV13 use in children would cost 561,682 USD per quality-adjusted life year (QALY) gained, compared to 62,065 USD per QALY estimated in 2013 due to declines in remaining vaccine-preventable disease burden in adults.⁹ However, ACIP recognized that some adults aged ≥65 years who are at increased risk for exposure to PCV13 serotypes or at increased risk for pneumococcal disease as a result of underlying medical conditions may benefit from PCV13 vaccination, and recommended PCV13 use based on shared clinical decision making.¹⁰ Adults with underlying conditions continue to be at increased risk for pneumococcal disease: a recent study reported that adults with CMC can have 2.3–15.4 times the incidence of IPD compared to adults without indication for pneumococcal vaccination.³ ACIP does not currently recommend PCV13 to adults with CMC, and PPSV23 alone is recommended for routine use in adults aged ≥19 years with CMC. Thus, we sought to evaluate the cost-effectiveness of implementing PCV13 in series with PPSV23 for all adults aged ≥19 years with CMC.

Methods

Model

In 2019, we developed a probabilistic model to estimate the public health impact and the cost-effectiveness of administering a dose of PCV13 at the diagnosis of CMC (chronic heart disease, chronic lung disease, diabetes, alcohol abuse, or chronic liver disease) to a hypothetical cohort of 19 year old adults in the United States. We used Monte Carlo simulation via spreadsheet-based software¹¹ to estimate the impact of administering PCV13 in series with PPSV23 on program costs, medical costs, non-medical costs, and disease burden.

We tracked new disease incidence and deaths due to IPD and non-bacteremic pneumococcal pneumonia (NBPP) by single year of age through life expectancy or until age 100 years (Figure S1). We assumed that IPD results in hospitalization or death, and NBPP results in either outpatient visit, hospitalization, or death. We estimated hospitalizations due to IPD as well as inpatient and outpatient encounters due to NBPP by age group (age 19–64 years, ≥65 years). We modeled the effects of replacing the currently recommended schedule of administering PPSV23 alone at CMC diagnosis with a schedule of administering PCV13 at CMC diagnosis followed by PPSV23 a year later. Adults with CMC receive only one lifetime dose of PCV13, but if PPSV23 was administered before age 65 years, a second dose is administered at age 65 years or later so that at least 5 years has lapsed since the last PPSV23 dose.¹² Adults who developed CMC after age 65 years would receive PPSV23 at age 65 years and a dose of PCV13 at the time of CMC diagnosis. We used the societal perspective for this analysis.

Study population

We used a cohort of 4,256,608 19-year-olds in 2017 in the United States. We stratified the population by single year of age and followed the population from age 19 through life expectancy or until age 100 years.¹³ We applied age-based incidence rates for individual CMC obtained from a variety of sources to the study cohort to estimate the size of CMC population by age. Incidence rates for adults with chronic heart disease and chronic liver disease were estimated from prevalence rates from the National Health and Nutrition Examination Survey,^14,15 and incidence rates for chronic lung disease, diabetes, and alcohol abuse were obtained or estimated from Centers for Disease Control and Prevention (CDC) surveillance systems.^16,17

Model parameters

Disease burden estimates, vaccine effectiveness and coverage, and cost per case used as model inputs are summarized in Table 1. For multivariate sensitivity analyses, we draw from lognormal distributions for cost parameters, pert distributions for vaccine coverage and effectiveness parameters, and normal distributions for all other model inputs.

Table

All-cause Pneumonia	Age groups
	19–44 years		45–64 years		65–74 years		75–84 years		≥85 years
	Value	95% CI	Value	95% CI	Value	95% CI	Value	95% CI	Value	95% CI
Inpatient All-cause Pneumonia Rate (per 100,000)^c	200.25	(150.0; 250.5)	579.85	(469.7; 690.0)	1216.10	(792.8; 1639.4)	2233.20	(2067.7; 2398.7)	3182.9	(2067.7; 4298.1)
Case fatality ratio (%)^d	2.4	(0.8; 3.9)	3.2	(1.2; 5.1)	3.7	(1.3; 6.1)	4.7	(2.1; 7.4)	7.2	(4.9; 9.4)
Outpatient All-cause Pneumonia rate (per 100,000)^e	736.2		893.9		1338.1		2186.6		3165.7
PCV13 type (%)^f	5.3	(3.1; 7.5)	5.3	(3.1; 7.5)	5.1	(3.4; 6.8)	5.1	(3.4; 6.8)	5.1	(3.4; 6.8)

^cThe lower bound of all-cause pneumonia for all age groups come from Jain et al.¹⁸ The upper bounds for 19–64 year old age groups come from Hayes et al.¹⁹ The upper bounds for age groups ≥65 years come from Ramirez et al.²⁰ CMC specific rates were estimated using the ratio of disease among different risk groups described in Weycker et al²¹weighted by the prevalence of different population of the risk groups estimated from NHIS data/To obtain the base case we take the midpoint between the upper bound and lower bound.

^dSource: National Inpatient Sample 2015. The upper bound is the fatality rate for cases with ICD-9 code with 481 in any location. The lower bound is for cases with ICD-9 code 481 for primary location. The base case is the midpoint of these two bounds.

^eSource: Nelson et al.²²

^fSource: Isturiz et al.²³ Base case estimate for 19–64 year olds are from Oct 2015–2016 estimate for at-risk group. Since the %PCV13 type for the three observed years were essentially the same for this age group, we used the Oct 2015–2016 estimate for the low-risk group as the lower bound. Upper bound is created for the range to be symmetrical around the base value. Since risk-group specific values were not presented for adults aged ≥65 years, we used Oct 2013–Sept 2014 value as the base case estimate for this age group. For the lower bound, we use the estimate for Oct 2015–Sept 2016. Upper bound is created to be symmetrical around the base value.

Table

		Age groups
Vaccine effectiveness		19–64 years		≥65 years
Vaccine type	Outcome	Value	Range	Value	Range
PCV13	PCV13-type IPD (-ST3, +ST6C)^g	75	(41.4; 90.8)	67	(11; 88)
PCV13	ST3 IPD^h	0	(0; 45)	0	(0; 26)
PCV13	PCV13-type NBPP (-ST3)^i	45	(14.2; 65.3)	32.5	(3.9; 53)
PCV13	ST3 NBPP^j	0	(0; 45)	0	(0; 45)
PPSV23	PPSV23-type IPD^k	73	(56.0; 84.0)	67	(37; 73)
PPSV23	PPSV23-type NBPP^l	0	(0; 50)	0	(0; 50)

^gSource: Bonten et al.²⁴ for 19–64 year old. Pilishvili et al for age ≥65 years²⁵

^hSource for adults aged ≥65 years from Pilishvili et al.²⁵ For adults aged 19–64 year olds, we assumed that the upper range will be as high as what we estimated for ST3 NBPP.

ⁱSource: Bonten et al. for age 19–64 years²⁴ Suaya et al.²⁶

^jWe assume PCV13 ineffective against ST3 pneumonia based on results from serotype 3 IPD.²⁵ For the upper bound of effectiveness, we use the effectiveness of PCV13 against all vaccine-type pneumonia from Bonten et al.²⁴

^kSource: Falkenhorst et al.²⁷ For 19–64 year olds, pooled estimate from case-control studies was used. For ≥65 years old, we assumed the point estimate to be the same as PCV13.

^lSource: Schiffner-Rohe, Falkenhorst, Tin Tin Htar.^27–29

Table

	Age groups
	19–64 years		≥65 years
Vaccine coverage	Value	Range	Value	Range
PCV13 Coverage Rate (%)^m	24.5	(23; 26)	47.1	(46.4; 50.7)
PPSV23 Coverage Rate (%)^n	24.5	(23; 26)	61.3	(59.9; 62.7)

^mFor age 19–64 year olds, we assumed that PCV13 coverage will be as high as PPSV23 coverage. For adults aged ≥65 years, the lower bound is from Centers for Medicare and Medicaid Services data and the upper bound is from Pfizer IQVIA Claims data. The base case is the midpoint.

ⁿSource for 19–64-year-olds was National Health Interview Survey, 2017.³⁰ Source for adults aged ≥65 years was National Health Interview Survey 2014.³¹

Table

	Age groups
	19–64 years		≥65 years
Cost per Case, 2017$^o	Value	95% CI	Value	95% CI
IPD	45,146	(35,915; 55,178)	30,459	(24,709; 36,753)
Inpatient NBPP	39,285	(28,628; 51,841)	26,187	(19,015; 35,630)
Outpatient NBPP	143	(86; 246)	285	(136; 519)

^oSource: CMS Medicare Claims from 2010–2015 inflated by Consumer Price Index for all items to 2017$.

Table

Cost associated with vaccination, 2017$	Value
PCV13^p	192.64
PPSV23^p	98.85
Vaccine administration cost per vaccination^q	26.61
Patient time and travel cost per vaccination^r	30.12

CI: confidence interval; CMC: chronic medical condition (chronic heart disease, chronic lung disease, diabetes, alcohol abuse, or chronic liver disease); IPD: invasive pneumococcal disease; NBPP: non-bacteremic pneumococcal pneumonia; PCV13: 13-valent pneumococcal conjugate vaccine; PPSV11: 11 serotypes unique to PPSV23 but not in PCV13 (STs 2, 8, 9 N, 10A, 11A, 12 F, 15B, 17 F, 20, 22 F, 33 F); PPSV23: 23-valent pneumococcal polysaccharide vaccine; ST: serotype

^p95% of the average wholesale price as reported by Medicare Part B, December 2017.

^qEstimated from the Medicare reimbursement for immunization administration (HCPCS code 90471) in 2017 averaged across all Medicare Administrative Contractors

^rSource: Maciosek et al.⁵² inflated by Consumer Price Index for all items to 2017$.

Table 1. Model input parameters for U.S. adults with chronic medical conditions, by age group

	Age groups
	19–64 years		≥65 years
IPD^a	Value	95% CI	Value	95% CI
Rate (per 100,000 population)	16.25	(15.27; 17.29)	40.89	(38.17; 43.80)
Case fatality ratio (%)	12.40%		15.71%
PCV13 (+6 C-3-19 F) type (%)^b	11.80%		7.20%
ST3 (%)	14.74%		16.22%
ST19F (%)	3.00%		1.90%
PPSV11 (%)	43.66%		38.41%

^aSource: ABCs data 2016–2017. Note inputs from ABCs are displayed here for reproducibility of final results. The number of digits displayed should not necessarily be construed as a measure of accuracy of these outputs from ABCs data. Population denominators to construct rates come from NHIS.

^bWe group serotype 6 C with PCV13 types due to cross-reactivity with serotype 6A. We treat serotypes 3 and 19 F separately due to observed performance in ABCs data.

Disease parameters

Pneumococcal disease was classified by serotype (ST) groups based on available information on direct PCV13 or PPSV23 effects among adults or indirect effects on adult disease from the childhood PCV13 program. Due to lack of further population-level impact of PCV13 on disease caused by ST3 and 19 F,^32–35 ST 3 and 19 F were tracked separately from other PCV13 types (1, 4, 5, 6A, 6B, 7 F, 9 V, 14, 18 C, 19A, and 23 F) for indirect effects, and ST 3 was tracked separately for direct effects (see “Vaccine effectiveness and coverage” below). ST6C was included as part of PCV13 types given the cross-reaction with ST6A.^36,37 STs unique to PPSV23 included the 11 STs contained in PPSV23 but not in PCV13 (ST2, 8, 9 N, 10A, 11A, 12 F, 15B, 17 F, 20, 22 F, 33 F). In 2016–2017, the proportion of ST3 IPD alone constituted 14.7% of all IPD in adults aged 19–64 years with CMC, and 16.2% in adults aged ≥65 years with CMC (Table 1).

The number of ST-group-specific IPD cases among adults 19–64 years old and ≥65 years old with any of the CMC conditions were obtained from CDC’s Active Bacterial Core surveillance (ABCs) during 2016–2017 (CDC unpublished data). To calculate IPD incidence, we used ABCs case counts as numerators and age-group specific population estimates for persons with CMC obtained from the 2016–2017 National Health Interview Survey (NHIS) as denominators.³⁸ Case fatality ratios were also obtained from ABCs by age group (Table 1).

For NBPP, we first obtained age-group specific incidence rates for all-cause pneumonia (including healthy, with CMC, and immunocompromised) from literature.^18–20,22 We then estimated age-group-specific all-cause pneumonia incidence for adults with CMC using incidence ratios comparing all-cause pneumonia rates among adults with CMC to that among healthy adults from the literature,²¹ and corresponding population denominators from NHIS. We used data from a multi-site observational study of adults hospitalized with pneumonia to estimate the proportion of all-cause pneumonia due to PCV13 ST.²³ The proportion of all-cause pneumonia caused by ST unique to PPSV23, ST 3 and 19 F, was estimated by applying the ST distribution for IPD based on ABCs data. Case fatality ratios for inpatient NBPP were obtained from the 2014 National Inpatient Sample.³⁹ We assumed that there were no deaths associated with outpatient NBPP.

We assumed that PCV13 type disease (including ST6C, excluding ST3 and ST19F) had an annual disease reduction of 4.1% through indirect effects from the childhood PCV13 program.³² We assumed that there were no further declines of pneumococcal diseases caused by ST3 and ST19F through indirect effects. Given that pneumococcal carriage in children is considered to be the main source of transmission of pneumococci to adults,⁴⁰ and given the low vaccine-type pneumococcal carriage in adults post PCV introduction in children,^41–43 we assumed that vaccinating adults with PCV13 would not result in additional indirect effects. In addition, serotype replacement has not been observed to date in the United States post PCV13-introduction in children.^44,45 Therefore, we did not consider serotype replacement in our model.

Vaccine effectiveness and coverage

Vaccine effectiveness (VE) was assumed to be higher for adults with CMC aged 19–64 years compared to adults aged ≥65 years. For adults with CMC aged ≥65 years, the VE estimate of PCV13 against PCV13-type (including ST6C, excluding ST3) IPD was based on the results of a U.S. case-control study.²⁵ VE estimate of PCV13 against PCV13-type NBPP came from a post-hoc analysis of the Community-Acquired Pneumonia Immunization Trial in Adults (CAPITA).²⁶ Given that no impact of PCV13 on ST3 IPD was observed at the population level since PCV13 introduction for adults in 2014¹⁸ and PCV13 was not effective against ST3 IPD in a U.S. post-licensure study,²⁵ we assumed zero as the point estimate and lower bound for PCV13 VE against both ST3 IPD and NBPP. For VE estimate of PPSV23 against IPD, we assumed the base case to be the same as PCV13, and used the point estimates of a meta-analysis for the upper and lower range.²⁷ Our base case assumed that PPSV23 was not effective against NBPP based on results of several meta-analyses (Table 1).^27–29 For adults aged 19–64 years with CMC, VE estimates of PCV13 against PCV13-type IPD and NBPP were based on the CAPITA data.²⁴ PPSV23 VE against PPSV23-type IPD was obtained from a meta-analysis.²⁷

Given the uncertainties around the estimates for PCV13 VE against ST3 pneumococcal disease and a range of estimates obtained from different post-licensure studies, we performed one-way sensitivity analyses where PCV13 VE against ST3 disease was estimated to be as high as VE against all-PCV13-type disease (75% for IPD and 45% for NBPP for 19–64 year olds,²⁴ 67% for IPD and 32.5% for adults aged ≥65 years²⁶).

We assumed no waning of immunity for 5 years after PCV13 receipt;⁴⁶ we assumed that the VE waned by 10% every 5 years thereafter until age 70. We assumed linear declines within 5-year increments. From age 70 to 85, we assumed a linear decline from the remaining VE to zero.⁴⁷ Because VE was assumed to be zero for those aged ≥85 years, we assumed that adults who develop CMC at age ≥85 years will not receive PCV13. Consistent with previous analyses,⁹ we assumed that after receipt of PPSV23, there was a linear decline to 50% from the original VE by year 5, then to 30% by year 10, then to 0% by year 15. If the patient is revaccinated at or after age 65 years (5 years after the last PPSV23 vaccination), then we assumed that the full VE (67%) against PPSV23-type IPD was achieved in year 1 and followed the same waning pattern over the next 15 years. Regardless of the order of sequence in which PCV13 and PPSV23 were given, we applied the VE estimate that was higher for STs that were common between PCV13 and PPSV23.

We obtained PPSV23 coverage estimates from 2017 NHIS data for adults aged 19–64 years at increased risk for pneumococcal disease³⁰ and 2014 NHIS data for adults aged ≥65 years. We used 2014 coverage estimates for older adults since 2015–2017 NHIS does not distinguish between PCV13 and PPSV23 administration.³¹ For adults aged 19–64 years, we assumed that PCV13 coverage will be as high as PPSV23 coverage. For adults aged ≥65 years, we used coverage data from Medical Claims as of June 2017 for lower bound,⁴⁸ and IQVIA claims data as of September 2017 for the upper bound.^49,50 We used the midpoint of these estimates as the point estimate for PCV13 coverage.

Health utility and medical and non-medical costs

We applied the same assumptions for health utility and medical and non-medical costs as in our previous study.⁹ Briefly, we used QALY decrements to account for differences in disease severity across pneumococcal disease (e.g., IPD, NBPP, disease treated as inpatient and outpatient). We used 0.00865 for IPD, 0.006 for inpatient NBPP, and 0.004 for outpatient NBPP.⁵¹ Each of these QALY decrements were scaled by age-specific baseline QALY values for adults with CMC ranging from 0.51 for those aged ≥86 years to 0.72 for adults aged 19 years. Cost per episode of pneumococcal disease was calculated from Centers for Medicare and Medicaid (CMS) Medicare claims from 2000 to 2015 (Table 1). We used 95% of the average wholesale price for PCV13 ($192.64) and PPSV23 ($98.85) as reported by Medicare Part B (December 2017) (Table 1). We used 26.61 USD for the cost of vaccine administration, estimated from the Medicare reimbursement for immunization administration (HCPCS code 90471) in 2017 averaged across all Medicare Administrative Contractors. We estimated patient time and travel cost at 30.12 USD.⁵² We inflated costs to 2017$ using the Personal Consumption Expenditure deflator for health accounts for medical costs and the Consumer Price Index for non-medical costs,⁵³ and discounted all future outcomes and costs by 3% per year.

Multivariate sensitivity analysis

In addition to the one-way sensitivity analyses mentioned above, we conducted multivariate sensitivity analysis using ranges for model inputs indicated in Table 1 using previously described methods.⁹

Results

Under base-case assumptions (Table 1), adding a dose of PCV13 for adults with CMC resulted in 54 fewer IPD cases, 319 fewer hospitalized NBPP cases, 565 fewer non-hospitalized NBPP cases, 4 fewer IPD deaths, and 10 fewer NBPP deaths over a lifetime of a single cohort (Table 2). This is predicted to save a total of 174 discounted QALY in this cohort over the entire follow-up period with a total discounted cost of 120 USD million, resulting in a cost of 689,299 USD per QALY (5^th percentile: 165,232 USD per QALY; 95^th percentile: 686,856 USD per QALY). In the one-way sensitivity analyses, the cost per QALY declined to 79,416 USD when we assumed that PCV13 VE against ST3 disease was equivalent to the VE estimates for other PCV13 STs. The decline was more substantial if we assumed higher VE estimates against ST3 NBPP ($93,184 per QALY gained) compared to ST3 IPD ($431,419 per QALY gained) (Table 3).

Table 2. Impact of adding PCV13 at diagnosis of chronic medical conditions, base case

	No PCV13 (A)	Adding PCV13 (B)	Base Case (B)-(A)	(5^th percentile; 95^th percentile)
Health Outcomes
IPD Cases	9,121	9,068	−54	(−77; −37)
Hospitalized NBPP Cases	144,896	144,577	−319	(−1,063; −216)
Non-hospitalized NBPP Cases	200,824	200,258	−565	(−1,678; −381)
Deaths due to IPD	880	876	−4	(−6; −3)
Deaths due to NBPP	5,304	5,293	−10	(−37; −6)
Discounted QALYs gained	−53,314	−53,141	174	(176; 614)
Discounted life-years gained	−83,087	−82,831	255	(198; 646)
Costs (million $)
Total Cost	3,426	3,546	$120	(99; 124)
Medical Costs	3,254	3,243	-$11	(−31; −9)
Vaccine Costs	172	303	$131	(126; 135)
Cost Ratios ($)
Cost/QALY	–	–	689,299	(165,232; 686,856)
Cost/Life-year	–	–	468,449	(156,621; 611,947)

IPD: invasive pneumococcal disease; NBPP: non-bacteremic pneumococcal pneumonia; QALY: quality-adjusted life year

Table 3. Impact of adding PCV13 at diagnosis of chronic medical conditions, one-way sensitivity analyses

	PCV13 VE vs ST3 IPD Equal to Protection Against other Serotypes of IPD^a	PCV13 VE vs ST3 NBPP Equal to Protection Against other Serotypes of NBPP^a	PCV13 VE vs ST3 NBPP and IPD Equal to Protection Against other Serotypes of Disease^a
Health Outcomes
IPD Cases	−141	−54	−141
Hospitalized NBPP Cases	−319	−2,244	−2,244
Non-hospitalized NBPP Cases	−565	−3,427	−3,427
Deaths due to IPD	−12	−4	−12
Deaths due to NBPP	−10	−77	−77
Discounted QALYs gained	269	809	904
Discounted life-years gained	393	1,243	1,382
Costs (million $)
Total Cost	$116	$75	$72
Medical Costs	-$15	-$55	-$59
Vaccine Costs	$131	$131	$131
Cost Ratios ($)
Cost/QALY	431,419	93,184	79,416
Cost/Life-year	294,922	60,616	51,981

IPD: invasive pneumococcal disease; NBPP: non-bacteremic pneumococcal pneumonia; PCV13: 13-valent pneumococcal conjugate vaccine; QALY: quality-adjusted life year; ST: serotype

^aWhen PCV13 is assigned equal VE against ST3 disease as against other serotypes it is assigned 75% vs IPD and 45% vs NBPP for the 19–64 age group and 67% vs IPD and 32.5% vs NBPP for the 65+ age group

A multivariate sensitivity analyses on all model parameters showed that PCV13 VE against ST3 NBPP in adults aged 19–64 years old had the greatest impact on cost per QALY: when parameter values were in the top 10% of those drawn from the range in Table 1, the mean cost per QALY was 244,600 USD; the cost per QALY was 493,059 USD when parameter values were in the bottom 10% (Figure 1). Indirect effects from the childhood PCV13 program and PCV13 VE against ST3 NBPP in adults aged ≥65 years had the next largest impact on the cost-effectiveness ratio. As the ACIP does not use a cost-effectiveness threshold, Figure S2 shows the cumulative density of cost per QALY estimates returned by the model across the full range of thresholds. In 99.39% of simulations the cost was more than 100,000 USD per QALY, in 89.96% of simulations the cost was more than 200,000 USD per QALY, and 50% of simulations had costs higher than 350,000 USD per QALY.

Figure 1. Tornado diagram of model sensitivity to outputs

CMC: chronic medical condition (chronic heart disease, chronic lung disease, diabetes, alcohol abuse, or chronic liver disease); IPT: inpatient; PCV13: 13-valent pneumococcal conjugate vaccine; pneumonia: non-bacteremic pneumococcal pneumonia; PPSV23: 23-valent pneumococcal polysaccharide vaccine; QALY: quality-adjusted life year; ST: serotype

Discussion

Our results showed that adding a dose of PCV13 to the currently recommended PPSV23 for adults with CMC will provide limited public health impact against preventing pneumococcal disease under base case assumptions for our model inputs. The estimated impact on disease outcomes was smaller, with a larger cost per QALY gained than what was recently estimated for vaccinating all immunocompetent adults (including those with CMC conditions) aged ≥65 years.⁹ This is likely because although most of the disease burden occurs in older adults, the transition to vaccination of persons with CMC happens earlier in life when disease risk may be lower.

As shown by the tornado diagram ur results were sensitive to the assumptions around PCV13 effectiveness against ST3 disease, particularly NBPP. In our base case, we assumed that PCV13 had no effectiveness against ST3 disease. Post-licensure studies report varying estimates of PCV13 effectiveness against ST3 disease. A recent meta-analysis based on three studies estimated PCV13 VE against ST3 hospitalized CAP 52.5% (95% CI: 6.2–75.9%), which is higher than the VE estimated from a randomized-controlled study for all vaccine-type pneumococcal pneumonia.^54,55 However, immunogenicity studies in infants^56–60 and adults⁵⁵ have generally shown a lower immunologic response to ST3 compared to the immune response to other PCV13 ST. In addition, population-level surveillance data have not demonstrated a PCV13 impact on ST3 disease in the U.S. after PCV13 introduction among adults aged ≥65 years. Surveillance for hospitalized pneumonia among adults in 21 U.S. hospitals showed that although the proportion of overall PCV13-type pneumonia decreased 2 years after PCV13 introduction for adults, the proportion of ST3 pneumonia increased in adults ≥65 years.²³ ABCs data showed that during 2014–2018, ST3 IPD incidence increased during a period when PCV13 coverage reached nearly 50% while IPD caused by other PCV13 STs decreased among adults aged ≥65 years (ABCs unpublished data),³² for whom routine PCV13 vaccination was recommended in 2014. Given these data, the base case assumption in our model of no PCV13 effectiveness against ST3 disease was consistent with lack of population-level impact on ST3 disease in the U.S.

Our study is subject to several limitations. First, data on VE of PPSV23 against vaccine-type pneumococcal pneumonia are inconsistent, leading to uncertainties of this estimate; many studies, including meta-analyses, have demonstrated no effectiveness. As in our previous studies among older adults,^9,61 assuming effectiveness of PPSV23 against pneumococcal pneumonia 45% (equivalent to PCV13 VE²⁴) would have resulted in higher cost per QALY gained for PCV13. Second, we assumed that PCV13 coverage in adults aged 19–64 years will be the same as PPSV23 coverage, which is much lower than that among adults aged ≥65 years. However, our sensitivity analysis showed that PCV13 coverage did not have significant impact on the outcome of the model. Third, we have limited data on the effectiveness or duration of protection of the vaccines in people with CMC, especially in younger adults. Additionally, people with CMC consist of groups with varying degree of risk against pneumococcal disease.³ We attempted to address the uncertainty through multivariate sensitivity analyses, which showed that the results were most sensitive to the assumption of PCV13 VE against ST3 pneumonia. Most of the studies we referenced targeted older adults; therefore, it is possible that we underestimated VE or duration of protection from PCV13 in younger adults with CMC. Lastly, there are limited data on utility decrements to use for pneumococcal diseases. We used values from published reports;⁵¹ however, higher utility decrements have been use in other models, which, if applied, would have reduced the cost per QALY gained for our base case.⁶²

In summary, our study showed that the strategy to administer PCV13 to adults aged ≥19 years with CMC is not economically favorable and is expected to lead to limited public health benefits due to reductions in disease burden through pediatric PCV13 use and small residual vaccine-preventable disease burden among adults with CMC. The strategy with PCV13 for adults with CMC had higher costs per QALY gained than recommending routine immunization for all immunocompetent adults aged ≥65 years.⁹ Our results were sensitive to the assumptions around PCV13 effectiveness against ST3 disease, particularly NBPP. Higher-valency pneumococcal conjugate vaccines covering broader range of serotypes causing disease among adults are expected to be licensed in the near future,^63,64 and will provide opportunities to reduce remaining pneumococcal disease burden in adults with CMC including ST3 disease if these new vaccines consistently demonstrate population-level impact against ST3 disease.

Acknowledgments

Dr. Stoecker’s work on this project was funded by an Intergovernmental Personnel Agreement with the Centers for Disease Control and Prevention (CDC).

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Additional information

Funding

This work was supported by the Centers for Disease Control and Prevention.