The US Food and Drug Administration (FDA) licensed 23-valent pneumococcal polysaccharide vaccine (PPSV23) (Pneumovax^®23, Merck & Co, Inc., Whitehouse Station, NJ) in 1983 [1]. PPSV23 replaced 14-valent pneumococcal polysaccharide vaccine. Another licensed 23-valent pneumococcal polysaccharide vaccine, Pnu-Imune^®23 (Wyeth Lederle), was used from 1983 to 2002. Pneumovax^®23 is the only licensed pneumococcal polysaccharide vaccine currently available in the United States. The Advisory Committee on Immunization Practices (ACIP) recommends PPSV23 for individuals aged 2 through 64 years who are immunocompetent with chronic conditions, have functional or anatomic asplenia or are immunocompromised; and for all adults aged >65 years regardless of previous medical history [2–7]. Although no human or animal pre-licensure data are available on the safety of PPSV23 in pregnancy [1], pregnancy is neither a contraindication nor a precaution for vaccination [8]. In pre-licensure trials, the most frequently reported adverse events (AEs) after PPSV23 included injection site reactions like pain, swelling and redness at the injection site, and headache, fatigue, and myalgia. Injection site reactions were higher in persons receiving a second dose of the vaccine compared to persons receiving the first dose – pain was observed in 60% of primary vaccinations versus 77% of revaccinations, injection site swelling or induration was reported in 20% versus 40%, and injection site erythema was reported in 16% versus 35% respectively. Serious AEs were rare [1].

PPSV23 has been used for decades and has an extensive clinical record. However, limited post-licensure safety assessment has been conducted [9–11]. We analyzed reports submitted to the Vaccine Adverse Event Reporting System (VAERS) following PPSV23 for the period 1990 through 2013.

VAERS is a national vaccine safety monitoring system jointly administered by the Centers for Disease Control and Prevention (CDC) and the FDA that accepts spontaneous reports of AEs following vaccination [12]. VAERS accepts reports from patients, parents, healthcare providers, vaccine manufacturers, and others. The VAERS report form collects information on the patient, vaccines administered and the AE experienced. Signs and symptoms of AEs are coded using the Medical Dictionary for Regulatory Activities (MedDRA), a clinically validated, internationally standardized terminology [13]. VAERS may also receive reports of vaccination errors (e.g., incorrect dose administered) not describing an AE. A VAERS report may be assigned one or more MedDRA preferred terms (PT). Reports are classified as serious based on the Code of Federal Regulations if one or more of the following is reported: death, life-threatening illness, hospitalization or prolongation of existing hospitalization, or permanent disability [14]. For non-manufacturer serious reports, medical records are routinely requested and made available to VAERS personnel; for reports of deaths, efforts are made to obtain autopsy reports and death certificates to ascertain cause of death.

We analyzed the safety of PPSV23 using four methods: (1) automated analysis of VAERS data, including comparisons of PPSV23 reports and trivalent inactivated influenza vaccine (IIV3) reports, (2) crude AE reporting rates based on PPSV23 doses distributed in the US market for all reports, serious reports, and reports of cellulitis and anaphylaxis, (3) clinical review of death reports and reports involving vaccine administered to pregnant women, and (4) empirical Bayesian (EB) data mining to assess for disproportional reporting of any AE after PPSV23 compared to AEs after other vaccines in the VAERS database.

During the study period, VAERS received 25,168 PPSV23 reports (Table 1); 92% were non-serious, 67% were in females and 86% of were in adults aged ≥19 years. Healthcare providers submitted 42% of reports. The percentages of serious reports were higher in children aged 2–18 years compared to adults, but did not exceed 20% (Tables 2a and 2b). Based on an estimate of just over 108 million PPSV23 doses distributed in the United States from 1990 to 2013 (CDC unpublished data), the AE reporting rate for all reports was 23 per 100,000 doses distributed and for serious reports 2 per 100,000 doses distributed.

The most commonly reported AEs in children aged 2–18 years were pyrexia and various injection site reactions (Table 3). PPSV23 was given alone in 45% of reports in this age group. Among children who received other concomitant vaccines with PPSV23, 50% received IIV3, 10% received diphtheria, tetanus, and pertussis (DTaP) vaccine and 2% received monovalent 2009-H1N1 pandemic influenza vaccine. The most commonly reported AEs in adults aged ≥19 years were injection site erythema, injection site pain and pyrexia (Table 4). PPSV23 was given alone in 52% of reports in adults. Among adults who received other concomitant vaccines with PPSV23, 81% received IIV3.

When comparing PPSV23 reports to IIV3 reports when the vaccines were given alone, the percent of serious reports in children was greater for PPSV23 than for IIV3: in children aged 2–5 years, 20% versus 9%; 6–12 years, 20% versus 7%; 13–18 years, 18% versus 10%. However, in individuals aged ≥65 years, the percent of serious reports was greater for IIV3 compared to PPSV23,16% versus 8%.

In our VAERS review of PPSV23, we found that fever and injection site reactions were the most commonly reported AEs after PPSV23 in both children and adults. Cellulitis, an AE that has been observed in the post-marketing experience and is listed in the package insert [1,18,19], was a commonly reported serious AE. Cellulitis and cellulitis-like reactions can be difficult to distinguish from local reactions that involve pain, swelling, erythema, streaking and warmth, all of which occur after PPSV23. We did not detect any new or unexpected safety concerns in our analysis of PPSV23 reports.

Injection site reactions are known to occur more frequently after revaccination than after the initial dose [1,19]. In our data, 43% of PPSV23 reports did not include dose number. Prior experience also indicates that dose number is inconsistently reported on the VAERS form for many vaccines. Because of these limitations, we were not able to assess injection site symptoms by PPSV23 dose number. PPSV23 can be administered by intramuscular or subcutaneous injection [1]. There is evidence to suggest that the intramuscular route is associated with less injection site reactions than the subcutaneous route [20–22]. However, we are not able to compare AEs for intramuscular versus subcutaneous PPSV23 administration in VAERS due to data completeness and quality issues.

The AE reporting rate for all PPSV23 reports was 23 per 100,000 doses distributed and for serious reports 2 per 100,000 doses distributed. This is higher than that observed in a review of IIV3 reports to VAERS from 1990 to 2005, which showed a reporting rate of 2.44 per 100,000 doses administered for all reports and for serious reports 0.4 per 100,000 [23]. To further evaluate differences between PPSV23 and IIV3 reports, we conducted post hoc analysis that showed when the vaccines were administered alone, the percent of serious reports in children was greater for PPSV23 compared to IIV3, but in individuals aged ≥65 years the percent of serious reports was greater for IIV3 compared to PPSV23. When considering the recommendations for PPSV23 [2–7], these findings might be due to the epidemiologic phenomenon of confounding by indication [24]. In children, PPSV23 is recommended for those with certain chronic illnesses and functional or anatomic asplenia (e.g., sickle cell disease, splenectomy); illnesses that are generally risk factors for significant adverse health events. In contrast, annual seasonal influenza vaccination has been recommended for all children regardless of health status [25]. For most of the study period, in the elderly PPSV23 had generally been recommended as a onetime single dose for all individuals at age 65 years [26], while annual seasonal influenza vaccination has been recommended for all individuals aged ≥65 years [25]. The post hoc analysis also revealed differences between the two vaccines in the distribution of reports within the ≥65 years old age group, with the percent distribution of IIV3 reports weighted toward older individuals within this elderly age group compared to PPSV23 reports, which conversely were weighted toward younger individuals. Therefore, differences in percentage of serious reports between PPSV23 and IIV3 might best be explained by the different indications and recommendations for these vaccines.

We did not compare data for PPSV23 with the pneumococcal conjugate vaccine (PCV) in VAERS. During the study period, 1990–2013, PPSV23 and PCV were approved and recommended for very different groups [2–4,26]. A comparison between PPSV23 and PCV vaccines in VAERS would be confounded by indication and difficult to interpret.

Our specific reviews of anaphylaxis, pregnancy, and deaths found no concerning patterns. Our reporting rate of anaphylaxis after PPSV23, 0.6 cases per million doses of vaccine distributed, is lower than a published estimate of approximately 1 case per million doses of any type of vaccine administered to children and adolescents [27]. This is reassuring; however, under-reporting in VAERS likely explains the lower reporting rate observed [28]. Although PPSV23 is not specifically recommended for pregnant women, pregnancy is neither a contraindication nor a precaution [8]; only 17 pregnancy-related reports were submitted to VAERS during the study period. Most of these reported non-pregnancy-specific AEs, such as injection site reactions, or no AE; there were only two reports of fetal demise, both spontaneous abortions. For the four deaths reported in children, non-vaccine-related causes were listed in the autopsy reports, death certificates, and/or medical records. Clinical reviews of death reports did not reveal any concerning patterns that would suggest a causal association with PPSV23 in children or adults. One death was attributed to anaphylaxis following PPSV23, however, medical records indicate the patient had a complicated hospitalization with a diagnosis of metastatic renal cell carcinoma and major abdominal surgeries and died in the hospital shortly after receiving PPSV23. Our finding that few deaths were reported after PPSV23, and the lack of causal association between vaccination and deaths, is consistent with a larger study of death reports submitted to VAERS following all vaccines. In that study, as in ours, causes of death were consistent with the most common causes of death in the US population [29].

Our study had several limitations. Although VAERS has broad national scope and the ability to rapidly detect safety signals and rare adverse events, it is subject to the limitations of spontaneous reporting systems, such as under-reporting, reporting biases, inconsistency in quality and completeness of reports, and lack of an unvaccinated comparison group [12]. A parent or patient may report the wrong vaccine (e.g., PPSV23 maybe reported when pneumococcal conjugate vaccine or another vaccine was received), or may report pneumococcal vaccine type unknown; in which case the vaccine was entered in VAERS as PPSV23. Both of these situations could lead to misclassification of the vaccine. Generally, we cannot determine if the reported AEs were causally related to receipt of PPSV23. Furthermore, estimates of crude reporting rates using doses of PPSV23 vaccine distributed as a denominator should be interpreted with caution, since the actual number of doses administered and to whom those doses were administered are not known, and the rate of underreporting is unknown. Despite these limitations, VAERS is a valuable monitoring system to detect potential vaccine safety problems that might require further investigations in more controlled studies. The present study detected no such concerns.

We did not identify any new or unexpected safety concerns in our study of PPSV23 reports to VAERS from 1990 through 2013. Our findings from VAERS data are consistent with findings from pre-licensure clinical trials [1] and other post-licensure studies [9–11,18,19].