The Effectiveness of 23-valent Pneumococcal Polysaccharide Vaccine on Elderly Colorectal Cancer Long-Term Survivors: A population-based exact-matched cohort study

ABSTRACT

Colorectal cancer (CRC) long-term survivor is a rapid enlarging group. However, the effectiveness of 23-valent pneumococcal polysaccharide vaccine (PPSV23) on this group is unknown. This nationwide population-based study in Taiwan was designed to examine the effect of PPSV23 on incidence rate ratio (IRR) of pneumonia hospitalization, cumulative incidence, and overall survival rate for these long-term CRC survivors. This cohort study was based on the Taiwan Cancer Registry and Taiwan National Health Insurance Research Database from 2000–2017. After individual exact matching to covariates with 1:1 ratio, there were a total of 1,355 vaccinated and 1,355 unvaccinated survivors. After adjusted by multivariate Poisson regression model, vaccinated group had a non-significantly lower pneumonia hospitalization risk than unvaccinated, with an adjusted IRR of 0.879 (p = .391). Besides, vaccinated group had both lower cumulative incidence rate and higher overall survival time than unvaccinated.

KEYWORDS:

• Pneumococcal polysaccharide vaccine
• colorectal cancer
• survivor
• pneumonia
• hospitalization

Introduction

Colorectal cancer (CRC) is a common cancer. According to Global Cancer Statistics 2020, CRC is the third most commonly diagnosed form of cancer globally, comprising 10% of all cancer diagnoses.¹ The prevalence of CRC is especially high in high human development index country.² In Taiwan, the CRC is also a major public health problem.³ According to the report of the Bureau of Health Promotion, Taiwan, the CRC has become the most common malignancy in Taiwan since 2006 with the standardized incidence rate of 42.9 per 100,000 people in 2019.⁴ Several early detection technologies of CRC, such as fecal occult blood test screening, and colonoscopy, improve survival time of CRC.^5,6 Besides early detection, advanced surgery method, like total mesorectal excision etc., also make CRC survivors become a growing population.^7,8

There were 43.8 million total cancer survivors who had been diagnosed within the past five years worldwide in 2018. Of them, 11% were CRC survivors.⁹ According to a report by American Cancer Society, majority of survivors are long-term survivors, living 5, 10 or more years after cancer diagnosis, and 47% of cancer survivors are 70 years old or older, and only 5% are younger than 40.^8,10

Infectious diseases account for high morbidity and mortality in patients with cancer. Pulmonary infection in the immunocompromised people is a major cause of morbidity and mortality, with mortality remaining high at 40% to 50%.¹¹ A previous study showed that pneumonia is associated with increased mortality, number and severity of complications, length of hospitalization, and hospital-related costs in cancer patients.¹² Among them, community-acquired pneumonia (CAP) represents the most common and life-threatening disease.^11,13,14

Streptococcus pneumoniae is the predominant pathogen in CAP in adults.¹⁵ It has been estimated that around 4000 (mostly adults) die in the United States each year because of S. pneumoniae.¹⁶ Patients with cancer are at a high risk for developing S. pneumoniae infection and further invasive pneumococcal disease.^17,18

For the effectiveness of 23-valent pneumococcal polysaccharide vaccine (PPSV23) to against S. pneumoniae in cancer patients, several studies have been performed. PPSV23 has been showed with adequate antibody response to S. pneumoniae in children with untreated Hodgkin’s disease.¹⁹ Our previous study also demonstrated that PPSV23 had well protective effectiveness in patients with lung cancer, even when PPSV23 was inoculated during active anti-cancer treatments period.²⁰ PPSV23 is also a very safe vaccine with adverse event following immunization reporting rate 12.4 × 100,000 doses and only 13.2% of them occurred in patients aged 60 or older.²¹

Among long-term cancer survivors receiving PPSV23 in Taiwan, the most common cancer type is colorectal cancer, accounting for up to 25% of all cancer survivors.²² The effectiveness of PPSV23 on CRC survivors has not been investigated before. Considering the growing CRC survivors population in the world, it is worth to investigate PPSV23 effectiveness on these specific cancer survivor type, CRC survivors. In this study, we focus and investigate on the effectiveness of PPSV23 on CRC long-term survivors, inoculated 5 years later after cancer diagnosis.

Materials and methods

Sources of data

The National Health Insurance Research Database (NHIRD) was provided by the Health and Welfare Data Science Center, released for research purposes by the National Health Insurance Administration, Ministry of Health and Welfare, Taiwan. The NHIRD contains medical data for approximately 99% of Taiwanese.²³ To ensure the accuracy of the claims, the National Health Insurance Administration (NHIA) performs quarterly expert reviews on ambulatory and inpatient claims filed.²⁴ False diagnostic reports are liable to penalties from the NHIA.²⁵ Information obtained from NHIRD is considered both complete and accurate.²⁶

The databases contained medical claims for ambulatory care, inpatient hospitalization, a registry of beneficiaries, and a registry of catastrophic illness certificates, from year 2000 to 2017. The registry of beneficiaries recorded socioeconomic data and the registry of catastrophic illness certificates were issued to all patients with pathologically confirmed malignant tumors to reduce the medical expense burden for cancer patients in Taiwan.

Patients and the study groups

Between 2000 and 2010, a total of 64,655 colorectal cancer patients were identified in Taiwan NHIRD and validated in the catastrophic illness registry. In Taiwan, the government started to execute a policy of free PPSV23 vaccination since year 2007 for people ≥75 years of age. At the time of year 2007, there were total 25,461 elder CRC patients diagnosed between year 2000 to 2007, and aged over 75 years old, qualified for free PPSV23 policy. Among them, there were 11,730 patients who received definite anti-cancer treatments rather than hospice care. Among the 11,730 CRC patients receiving definite treatments, there were total 3083 patients ever received the PPSV23 provided by the healthy promotion policy from year 2007 to 2017. We analyzed these above vaccinated CRC patients and found most patients (2,278 of 3, 083 patients, 73.9%) received PPSV23 in 2008 winter, October 2008 to December 2008 (Table 1).

Table 1. Month distribution of PPSV23 vaccination in elderly colorectal cancer patients.

	January ~ September		October ~ December
Years	Numbers	Percentage	Numbers	Percentage
2007	0	0.0%	287	9.3%
2008	10	0.3%	2278	73.9%
2009	6	0.2%	184	6.0%
2010	4	0.1%	104	3.4%
2011	3	0.1%	59	1.9%
2012	3	0.1%	36	1.2%
2013	0	0.0%	31	1.0%
2014	0	0.0%	17	0.6%
2015	0	0.0%	21	0.7%
2016	0	0.0%	23	0.8%
2017	0	0.0%	17	0.6%
2007~2017	26	0.8%	3057	99.2%
Total	3083		(100.0%)

Therefore, we defined the “vaccination period” as October 2008 to December 2008, which also reduced the potential immortal time bias associated with the competing risk by death, i.e., cancer patients dying too early to receive the vaccination. CRC patients who received PPSV23 outside the period 2008/10 ~ 2008/12 were excluded. CRC patients who died before January 1, 2009 or with other cancer history were also excluded (Figure 1). We excluded 1,086 patients died before 2009 and 805 patients who received PPSV23 outside the target vaccination period. After exclusion, there were total 7,562 elderly CRC patients never receiving PPSV23 and 2,277 patients receiving PPSV23 during 2008/10 ~ 2008/12.

Figure 1. Study design flowchart of cohort study for elderly colorectal cancer long-term survivors with and without PPSV23 vaccination. *Free PPSV23 policy in Taiwan started since year 2007 for people aged ≥75 years. ^#Month distribution of PPSV23 vaccination listed in table 1 showed most elderly colorectal cancer patients received PPSV23 within October to December, 2008 winter. The “vaccination period,” October 2008 to December 2008, is set to reduce immortal time bias and the observation time for study and control groups both started from 2009/1/1. **Cancer patients who survived at least 5 years by the time of year 2007, i.e., had cancer diagnosis before year 2002, were defined as long-term cancer survivors.

In this study, we investigate the vaccine effectiveness for CRC long-term survivor, we limited our CRC patients to those who has survived for at least 5 years, i.e., had CRC diagnosis before year 2002. There were total 4,574 elderly CRC long-term survivors never receiving PPSV23 and 1,403 survivors receiving PPSV23 during 2008/10 ~ 2008/12.

Considering the relatively low vaccination rate for elderly CRC patients (3,083 over 11,730; 26.3%), self-selection for vaccination may exist. To reduce the bias of confounding by indication that people with a history of frequent pneumonia would have a higher tendency to receive vaccination than general population, we performed individual exact matching, where each CRC survivors in the vaccination group gets one exact match from the non-vaccination group for these four covariates, age, gender, influenza vaccination, and number of previous pneumonia hospitalizations during past three years. After exact matching, each vaccinated CRC survivor had one matched unvaccinated survivor with the same age, gender, number of previous pneumonia hospitalizations during past three years, and the influenza vaccination status. After exact matching, there were total 1,355 vaccinated elderly CRC survivals and 1,355 matched unvaccinated counterparts.

Measurements of endpoints and potential confounders

The primary outcome in the study was hospitalization due to pneumonia (ICD-9-CM codes 481–482 and 485–486). Because we chose the period most CRC survivors received PPSV23, October 2008 to December 2008, as the defined “vaccination period,” the observation time in this study started from January 1, 2009, to the end of the study follow up, December 31, 2017, totally 9 years. Both invasive pneumonia and noninvasive pneumonia were included. Besides, we excluded viral pneumonia, pneumonia due to bacteria other than Streptococcus pneumoniae, and influenza. The confounders which were adjusted included age, gender, cancer treatment modalities, influenza vaccination, co-morbidity, and sociodemographic variables.

Charlson comorbidity index (CCI) score was adjusted in this study.²⁷ Comorbidities included in the CCI might impair immunity status, such as diabetes mellitus, liver cirrhosis, renal diseases, and connective tissue disease. Patients with other comorbidities in the CCI, like chronic obstructive pulmonary disease, and interstitial lung disease may had poor lung function and be predisposed to pneumonia. Other comorbidities which might also affect susceptibility to pneumonia, such as dementia, stroke, hemiplegia, coronary heart disease, and congestive heart failure, were all included in the CCI score calculation to be adjusted in the study.

Cancer treatments, which might affect cancer prognosis, such as surgery, radiotherapy, chemotherapy, and targeted therapy were also included into adjustment.^18,28,29

To reduce the bias of confounding by indication which people with higher health awareness might prefer to receive PPSV23 than the general population, we incorporated several sociodemographic variables, such as urbanization level, geographic region, and socioeconomic status, into the regression model to reduce their effects on the pneumonia hospitalization risk estimation. Urbanization level was adjusted because of the distinct urban – rural difference in medical care accessibility in Taiwan.³⁰ We grouped patients on the basis of urbanization level (i.e., urban, suburban, and rural) according to the classification scheme of Liu et al.³¹

Statistical analysis

We first performed exact matching for age, gender, influenza vaccination, and number of previous pneumonia hospitalizations during past three years. Patient characteristics between the two study groups were compared by a chi-square test. The pneumonia hospitalization incidence rate per 1000 person-years (PYs) was calculated with all pneumococcal pneumonia hospitalization episodes from January 2009 to December 2017 being counted. The incidence rate was adopted as the endpoint to reduce bias that different observation time among individuals, i.e. cancer patients died early due to any causes and had no or less chance to have a pneumonia. Because the dependent variable, the pneumococcal pneumonia hospitalization incidence, is a countable variable and thereby the incidence rate follows a Poisson distribution, we used a multivariate log-linear Poisson regression model to further calculate the incidence rate ratios (IRRs) with all covariates included. The overall survival time (from the start of the observation time, January 1, 2009, to the end of the study, December 31, 2017 or to the death) was compared by the Kaplan-Meier method. Two statistical packages (SAS [version 9.4; SAS Institute, Inc., Cary, NC and SPSS [version 12, SPSS Inc., Chicago, IL) were used to analyze the data. A two-sided P-value of < 0.05 was considered statistically significant.

Results

Baseline summary of demographic characteristics and co-morbidities for the two groups, PPSV23 vaccinated and PPSV23 unvaccinated long-term CRC survivors, was compared by the chi-square test and shown in Table 2. If p value of specific characteristic was less than 0.05, then its distribution between groups would be considered different. The characteristics between the groups were all similar (p > .05), except urbanization level (p = .002) and socioeconomic status (p = .036). The four factors, age, gender, influenza vaccination, and pneumonia hospitalization history all had exact same distributions (p = 1.000) after exact matching statistical method. PPSV23 vaccinated and unvaccinated elderly colorectal cancer long-term survivors had similar mean ± SD age, which was 80.8 ± 4.4 and 80.9 ± 4.6 years, respectively (p = .526 > 0.05). The CCI score distributions were also not different for these two groups (p = .144 > 0.05), as in the Table 2, compared by the chi-square test.

Table 2. Demographic characteristics and comorbidities of elderly long-term colorectal cancer survivors who had already survived for at least 5 years after cancer diagnosis, with and without PPSV23vaccination.

	Without PPSV23 Vaccination		With PPSV23 Vaccination
	N = 1,355		N = 1,355
Variables treatment	n	%	n	%	p-value
Age					–
75–79	601	44.4	601	44.4
80–84	501	37.0	501	37.0
≥85	253	18.7	253	18.7
Mean age (y/o ± SD)	80.9 ± 4.6		80.8 ± 4.4		.5261
Gender					–
Male	745	55.0	745	55.0
Female	610	45.0	610	45.0
Previous 3 years Pneumonia number					–
0	1,290	95.2	1,290	95.2
1	53	3.9	53	3.9
≥2	12	0.9	12	0.9
Influenza Vaccination					–
Yes	1,211	89.4	1,211	89.4
No	144	10.6	144	10.6
Surgery					.317
Yes	1,355	100.0	1,354	99.9
No	0	0.0	1	0.1
Chemotherapy					.672
Yes	24	1.8	27	2.0
No	1,331	98.2	1,328	98.0
Radiotherapy					.525
Yes	60	4.4	67	4.9
No	1,295	95.6	1,288	95.1
Target therapy					.654
Yes	2	0.1	3	0.2
No	1,353	99.9	1,352	99.8
Charlson Comorbidity Index					.144
0–1	211	15.6	211	15.6
2–3	365	26.9	309	22.8
4–5	263	19.4	274	20.2
6–7	229	16.9	244	18.0
≥8	287	21.2	317	23.4
Urbanization level					.002
Metropolis	377	27.8	296	21.8
Satellite cities	571	42.1	613	45.2
Rural areas	407	30.0	446	32.9
Geographic region					.085
North	734	54.2	768	56.7
Central	409	30.2	395	29.2
East and remote islands	15	1.1	26	1.9
South	197	14.5	166	12.3
Socioeconomic status					.036
Highest	294	21.7	290	21.4
High	24	1.8	8	0.6
Low	521	38.5	544	40.1
Lowest	516	38.1	513	37.9

The patient characteristic between the two study groups was compared by a chi-square test. Because the four factors were exact matched with exact same distributions between groups, their p values were marked as “─.”

A total 1,829 episodes of pneumonia hospitalization occurred over an observation period of 17,638 person-years (PYs) in 975 patients. The pneumonia hospitalization incidence rate was higher in vaccinated CRC survivors (104.5 per 1000 PYs [95% CI: 97.9–111.4) than unvaccinated survivors (102.8 per 1000 PYs; 95% CI: 96.2–109.8; Table 3), and a higher proportion of vaccinated than unvaccinated CRC survivors had no pneumonia hospitalization (64.3% vs. 63.8%; Table 4). On the other hand, proportions of CRC survivors who had > 0–1, >1–2, and > 2–3 pneumonia hospitalizations per PY were all consistently lower in the vaccinated than unvaccinated group (20.0% vs. 21.2%, 8.4% vs. 8.6%, and 3.3% vs. 3.4%, respectively).

Table 3. Incidence density of pneumonia hospitalization in elderly long-term colorectal cancer survivors with and without PPSV23vaccination.

PPSV23	PYs	No. of subjects with pneumonia hospitalization	No. of episodes of pneumonia hospitalization	ID* (per 1000 PYs)	95% CI
With PPSV23	8,984	484	939	104.5	97.9	111.4
Without PPSV23	8,654	491	890	102.8	96.2	109.8
Total	17,638	975	1829	103.7	99.0	108.6

*ID, incidence density, calculated as the rate of the number of pneumonia admission episodes per 1000 PYs. PPSV23, 23-valent Pneumococcal polysaccharide vaccine; PY, person-year; CI, confidence interval.

Table 4. Frequency distribution of pneumonia hospitalization episodes per year in elderly long-term colorectal cancer survivors with and without PPSV23vaccination.

	No. of pneumonia admissions per year
PPSV23	0	>0–1	>1–2	>2–3	>3
With PPSV23	871	271	114	44	55
%	64.3%	20.0%	8.4%	3.3%	4.1%
without PPSV23	864	287	116	46	42
%	63.8%	21.2%	8.6%	3.4%	3.1%
Total	1735	558	230	90	97
%	64.0%	20.6%	8.5%	3.3%	3.6%

PPSV23, 23-valent Pneumococcal polysaccharide vaccine.

Our analysis after adjustment for confounders showed that PPSV23 vaccination reduced pneumonia hospitalization risk but not significantly, with an IRR (incidence rate ratio) of 0.879 (p = .391; Table 5) in this CRC survivors cohort. CRC survivors who had old age would had high IRR, with IRR 1.911 for age 80–84 and IRR 2.493 for age 85 and above, compared to age 75–79. About gender, the adjusted IRR of 0.510 with p < .001 indicated that females were in significantly lower risk of hospitalized pneumonia than males.

Table 5. Unadjusted and adjusted incidence rate ratio of pneumonia hospitalization in elderly long-term colorectal cancer survivors with and without PPSV23vaccination.

	Unadjusted IRR				Adjusted IRR
	IRR	95% CI		p-value	IRR	95% CI		p-value
PPSV23 (Without, ref)	1				1
With	1.105	0.800	1.528	.545	0.879	0.655	1.180	.391
Age (75–79, ref)	1				1
80–84	1.722	1.206	2.458	.003	1.911	1.312	2.784	<.001
85+	1.892	1.183	3.025	.008	2.493	1.608	3.864	<.001
Gender (male, ref)	1				1
Female	0.474	0.318	0.708	<.001	0.510	0.346	0.752	<.001
Previous 3 years Pneumonia number (0, ref)	1				1
1	2.725	1.628	4.564	<.001	2.760	1.773	4.296	<.001
2	2.220	0.605	8.152	.229	7.369	3.808	14.263	<.001
Influenza vaccination (No, ref)	1				1
Yes	1.255	0.627	2.510	.521	1.886	0.798	4.459	.148
Surgery (No, ref)	1				1
Yes	–				–
Chemotherapy (No, ref)	1				1
Yes	1.049	0.186	5.920	.957	0.506	0.093	2.743	.430
Radiotherapy (No, ref)	1				1
Yes	1.026	0.459	2.291	.950	0.618	0.263	1.454	.270
Target therapy (No, ref)	1				1
Yes	–				–
Charlson Comorbidity Index (0–1, ref)	1				1
2–3	1.432	0.732	2.799	.294	1.592	0.789	3.212	.194
4–5	1.541	0.771	3.082	.221	1.800	0.882	3.673	.106
6–7	1.886	0.957	3.717	.067	1.703	0.831	3.487	.146
≥8	2.680	1.433	5.010	.002	3.662	1.880	7.129	<.001
Urbanization (Urban, ref)	1				1
Suburban	1.185	0.788	1.784	.415	1.390	0.946	2.040	.093
Rural	0.990	0.620	1.578	.965	0.928	0.568	1.516	.765
Geographic region (Northern, ref)	1				1
Central	1.049	0.733	1.503	.793	1.588	1.098	2.296	.014
Eastern	0.639	0.087	4.677	.659	0.170	0.017	1.743	.136
Southern	1.042	0.635	1.709	.871	1.386	0.842	2.282	.200
Socioeconomic status (Highest, ref)	1				1
High	0.719	0.054	9.658	.804	0.870	0.190	3.989	.858
Low	1.147	0.693	1.898	.594	0.909	0.530	1.558	.727
Lowest	1.442	0.893	2.326	.134	1.330	0.860	2.058	.200

PPSV23, 23-valent Pneumococcal polysaccharide vaccine; IRR, incidence rate ratio; Ref., reference; COPD, chronic obstructive pulmonary disease. Some statistical results did not converge because there were too few case numbers (marked as —).

CRC survivors who had high CCI score had high risk of pneumonia hospitalization. Influenza vaccination had no obvious effect on the risk of pneumonia hospitalization (p = .148). For CRC long-term survivors, nearly all subjects received definite surgical resection and nearly none received target therapy in both cohorts, therefore the adjusted IRR of surgery and target therapy did not converge and their effects on pneumonia hospitalization could not be evaluated in this long term CRC survivors cohort. The other two anti-cancer treatment modalities, chemotherapy and radiotherapy, did not have obvious effect on risk of pneumonia hospitalization in long-term survivors (p = .430, and p = .270).

For cumulative pneumonia hospitalization incidence rate, CRC survivors receiving PPSV23 had a non-significant lower rate than those who were not receiving PPSV23 (Figure 2, p = .322). PPSV23 group also had a non-significant higher overall survival rate than non-PPSV23 group (Figure 3, p = .123).

Figure 2. Comparison of cumulative incidence rate of pneumonia hospitalization between elderly colorectal cancer long-term survivors with and without PPSV23 vaccination (p = .322). Orange line = with PPSV23 vaccination, blue line = without PPSV23 vaccination.

Figure 3. Kaplan-Meier survival curves for elderly colorectal cancer long-term survivors with and without PPSV23 vaccination (p = .123). Orange line = with PPSV23 vaccination, blue line = without PPSV23 vaccination.

Discussion

Most previous PPSV23 studies in cancer patients were antibody response studies. For example, in a study in Norway, PPSV23 vaccination elicited adequately protective pneumococcal IgG antibody levels in both 35 cancer patients who had just received chemotherapy and 38 control patients who had not.³² Another study reported an adequate antibody response to PPSV23 in children with untreated Hodgkin’s disease.¹⁹

However, reviewing the literature, few studies ever investigated the effectiveness of PPSV23 on health outcomes in cancer patients. Unlike previous studies, we focus on evaluating the frequency of pneumonia hospitalization reduced by PPSV23.

With the financial support from a non-government organization in Taiwan, the Wang Jhan-Yang social welfare Foundation, PPV23 has been provided free of charge to all Taiwan residents with age ≥75 years since 2007.³³ Because winter is the peak season for respiratory disease infections in Taiwan, Taiwan Centers for Disease Control would implement the PPSV23 vaccination program between October to December every year to people who are eligible.³⁴ Using these data, we ever reported several PPSV23 vaccination effectiveness results in cancer patients, in those studies we followed up to two years, and focused on general diagnosed cancer patients. For example, we reported that PPSV23 inoculated during the active anti-cancer treatment period could effectively protect lung cancer patients from pneumonia (IRR = 0.74, p = .03).²⁰ Our team also reported that PPSV23 vaccination could protect general diagnosed CRC patients during two years observation time (IRR = 0.88, p = .04), while that study did not focus on CRC long-term survivors.³⁵

We ever reported benefit of PPSV23 (IRR = 0.695, p = .03) for long-term cancer survivors of all cancer type with only two years follow-up time.²² However, that study did not differentiate various cancer types, included some cancer types, such as stomach, lung and liver cancer, of which kind cancer patients might receive more immunosuppressive medications or treatments than other cancer types. Considering the growing population of CRC survivors, accompanied by advanced surgery technique and early cancer detection, it is worth to investigate the cost-benefit of PPSV23 vaccination to this major group of cancer survivors.

In this study with long follow up time from year 2009 to 2017, we analyzed the multivariate results, cumulative incidences, overall survivor rates. We found that the benefit of PPSV23 for CRC long-term survivors who survived from cancer for at least 5 years (adjusted IRR = 0.879, p = .391) was not very obvious compared to previous reported results for general diagnosed CRC patients (adjusted IRR = 0.88, p = .04).³⁵ One possible explanation is that CRC survivors may be not as immunocompromised as the general diagnosed CRC patients. Patients just receiving anti-cancer therapies were more susceptible to pneumonia than long-term survivors. Comparing our previous general CRC study and this CRC long-term survivors study, the PPSV23 inoculation timing should be considered as earlier as possible.

Pneumococcal conjugate vaccine (PCV) for children could has indirect effects to this elderly group. However, during our specific“vaccination period,” October 2008 to December 2008, during which PCV7 was not free of charge and PCV13 was not introduced to Taiwan yet. The major serotype, S. pneumoniae serotype 19A, in Taiwan was also not covered by the PCV7. Since our study focus on the population who received vaccine in the winter of 2008, an indirect effect of PCV vaccines to our study is not considered.

Even free of charge, we could find that from year 2007 to 2017, most, 73.9%, CRC long term survivor group received the free vaccination during the 2008 winter, the next year after this health promotion policy start. This health promotion welfare policy seems to be known by the general population for very short period. Taiwanese still lack of health awareness for the risk and severity of S. pneumoniae infection. Further encouraging means to promote vaccination rate and to raise health awareness, such as though social media may be needed.

Study strengths

First, this study was a nationwide population-based study using NHIRD, a database for reimbursement purpose of medical expenses, without nonresponse or possible loss of follow-up as community-based study may have, offered a good opportunity to explore the effectiveness of PPSV23 on CRC survivors. Second, this study used a person-years approach to determine incidence rate, reducing the bias from different observation time. Third, we adjusted co-morbidities with Charlson comorbidity index to reduce possible underestimate of the PPSV23 effectiveness, because immunocompromised people are more likely to receive vaccine than others. Fourth, this study adjusted sociodemographic variables to reduce confounding by indication, that is, the different health awareness among people.

Study limitations

First, this is an observational study, rather than a randomized study. Second, our data source is a health insurance database, not recording personal smoking history, degree of physical activity of patients, or body weight/body mass index. Third, because the health promotion policy “free PPSV23” is provided to people with age over 75 years in Taiwan, the conclusion of this study could only be applied to this age group.

Conclusions

Vaccinated CRC long-term survivors had a non-significantly lower pneumonia hospitalization risk than unvaccinated, with an adjusted IRR of 0.879 (p = .391). For cumulative incidence rate and survival time, vaccinated group had lower cumulative incidence rate and higher overall survival time than unvaccinated, but both did not reach statistically significant.

Although the clinical benefit for CRC long-term survivors did not reach statistically significant is in this study, CRC patients are still recommended to receive PPSV23 vaccine after cancer diagnosis.

Author contributions

Conceptualization: W.-Y. C., C.-Y. L.; Project administration: W.-Y. C., M.-S. L., H.-Y. L., S.-K. H.; Funding acquisition: W.-Y. C., M.-S. L; Resources: M.-S. L., S.-K. H.; Data curation: B.-H. Y., W.-Y. C.; Methodology: W.-Y. C., C.-Y. L.; Software: C.-Y. L.; Formal Analysis: B.-H. Y., W.-Y. C.; Writing – original draft: W.-Y. C.; Writing – review & editing: W.-Y. C., S.-Y. C., F.-C. H., C.-C. W., L.-C. C., C.-H. C.; Supervision: W.-Y. C. All authors have read and agreed to the published version of the manuscript.

Institutional review board statement

The Institutional Review Board of Dalin Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation (approval number, B10704014-2) approved this study and waived the requirement for informed consent from the people involved because the National Health Insurance Research Database (NHIRD) is a de-identified database and the data were analyzed anonymously.

Informed consent statement

Written informed consent for participation was not provided by the participants/legal guardians/next of kin because the NHIRD encrypts patients’ personal information to protect privacy and provides researchers with anonymous identification numbers associated with the relevant claims’ information, including sex, date of birth, medical services received, and prescriptions. Therefore, informed consent is not required to access the NHIRD.

Acknowledgments

This study is based on data from the National Health Insurance Research Database provided by the Health and Welfare Data Science Center, the National Health Insurance Administration, Ministry of Health and Welfare, Taiwan. However, the interpretation and conclusions contained herein are not those of the National Health Insurance Administration, Ministry of Health and Welfare, Taiwan.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets generated and/or analyzed during the current study are not publicly available in accordance with the policy of the Health and Welfare Data Science Center, Ministry of Health and Welfare, Taiwan, but are available from the corresponding author upon reasonable request.

Additional information

Funding

We appreciate the research grants from the Dalin Tzu Chi Hospital [DTCRD 106(2)-I-19] and the Buddhist Tzu Chi Medical Foundation [TCMF-A 108-06]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the article.