Subcutaneous vaccine administration – an outmoded practice

ABSTRACT

Subcutaneous vaccine (SC) administration is an outmoded practice which complicates vaccine administration recommendations. Local adverse events following immunization (AEFIs) are a recognized determinant of vaccine hesitancy/refusal which can lead to an increased prevalence of vaccine-preventable disease.

This extensive narrative review provides high-grade evidence that intramuscular (IM) administration of all vaccine types [adjuvanted, live virus and non-adjuvanted (inactivated whole cell, split cell and subunit)] significantly reduces the likelihood of local adverse events. This, combined with moderate grade evidence that IM injection generates significantly greater immune response compared with SC injection, allows a strong recommendation to be made for the IM injection of all vaccines except BCG and Rotavirus.

This will simplify vaccination practice, minimize the inadvertent misadministration of vaccines and potentially improve public trust in vaccination.

Introduction

Vaccination has made, and will continue to make, a very significant contribution to world health.¹ However, adverse events following immunization (AEFTs), including injection site reactions (ISRs), are a significant driver^2,3 of vaccine hesitancy and refusal. The latter has resulted⁴ in significantly increased risks of pertussis, varicella and pneumococcal infections in non-vaccinated children compared with vaccinated children.

Consequently, the definition and implementation of best vaccination practice (site, route and technique of injection) in terms of AEFIs (reactogenicity) and immune response (immunogenicity) are mandatory.

The current mantra⁵ for vaccination practice has been to administer adjuvanted vaccines by intramuscular injection, live virus vaccines by subcutaneous injection and non-adjuvanted, inactivated whole cell, split and subunit vaccines by either route. This complicated regimen for vaccine administration is due to the unacceptable reactogenicity⁶ of subcutaneously administered adjuvanted vaccines.

Evidence-based medicine (EBM) has been championed⁷ as a way of improving the quality of patient care through a stepwise process of formulating the clinical questions to be answered, collating and appraising relevant data and defining the optimal response.

The purpose of this review is to use EBM to seek to rationalize the route of administration of vaccines given by SC, IM or either routes. The PICO elements⁸ for this review are P = human vaccine recipients, I = intramuscular route of injection, C = subcutaneous route of injection and O = reactogenicity and immunogenicity of vaccines.

Methods

Searches were made using Pubmed. Google Scholar, Scopus, Embase, Biological Abstracts, Science Citation index, Cochrane Database of Systematic Reviews (CDSR), Cochrane Central Register of Controlled Trials (CENTRAL) and Databases of Abstracts of Reviews of Effects (DARE) using the following search terms and their word variants; “vaccines,” “administration,” “subcutaneous,” “intramuscular,” “adverse reactions” and “immunogenicity.” Manual searches were made from the following journals for the date in parenthesis to January 2020: Acta Paediatrica (1998), Acta Tropica (1980), American Journal of Medicine (1946), American Journal of Public Health (1971), American Journal of Tropical Medicine and Hygiene (1998), Annals of Internal Medicine (1995), Annals of Tropical Pediatrics (1999), Archives of Diseases of Childhood (1926), Bio Drugs (1998), Biologicals (1990), British Medical Journal (1991), Canadian Medical Association Journal (1911), Clinical Infectious Diseases (1999), Clinical and Vaccine Immunology (2006), European Journal of Pediatrics (1997), Infection and Immunity (1970), Journal of Pediatrics and Childhood (1998), Expert Review of Vaccines (2002), Human Vaccines (2005), Human Vaccines & immunotherapeutics (2012), Journal of Pediatrics (1995), Journal of Travel Medicine (1997), Journal of Tropical Pediatrics (1995), Lancet (1990), Medical Journal of Australia (2004), New England Journal of Medicine (1992), Pediatrics (1960), Pediatric Infectious Disease Journal (1995), Pediatrics International (1999), Public Health (1995), Scandinavian Journal of Infectious Disease (1997), Transactions of the Royal Society of Tropical Medicine and Hygiene (1920), Vaccine (1983) and to find additional studies where these were not abstracted.

Bibliographies of all relevant articles were searched for additional studies. All route comparative studies were included for analysis except those involving patients with chronic cutaneous, subcutaneous and muscular disorders and non-English language studies unless the full article was available for translation.

Results

Fifty-eight studies, which satisfied the inclusion criteria, were retrieved by the searches (51 by literature search, 7 by a manual search of appropriate Journals). They were divided into two study design groups, randomized trials and observational studies, as recommended in the GRADE guidelines.⁹ The former has the potential to provide moderate to high-grade evidence whilst the latter could only give very low to low-grade evidence.

Local reactogenicity data were recorded as warmth, pain, redness and swelling. These and immunogenicity data were collated into vaccine groups; adjuvanted vaccines, live virus vaccines and non-adjuvanted vaccines (inactivated whole cell, split cell and subunit). These are presented as Tables 1–3 respectively.

Table 1.

Adjuvanted vaccines and intramuscular compared with subcutaneous administration – reactogenicity and immunogenicity

Author	Study design	Patients	Intervention	Outcome
Wright et al10	Multi-center, randomized, double-blind, phase IV study.	Healthy US adults 18–61 y old n = 1564	Anthrax toxoid (AVA) vaccine administered according to 7 different protocols.	Reactogenicity IM < SC odds ratio for warmth, tenderness, erythema, induration, subcutaneous nodules. Immunogenicity IM not inferior to SC at 9 weeks post vaccination.
Marano et al11	Multi-center, randomized, double blind, phase IV study.	Healthy US adults. 18–64 y old n = 1005	Anthrax toxoid (AVA) vaccine administered according to 7 different protocols.	Reactogenicity IM < SC odds ratio for warmth, tenderness, erythema, induration, subcutaneous nodules and pain immediately after injection. Immunogenicity IM not inferior to SC administration.
Pittman et al12 and Pittman13	Single-center, randomized, double-blind study.	Healthy US adults 18–61 y old n = 173	Anthrax toxoid (AVA) vaccine administered according to 7 different protocols.	Reactogenicity SC > IM odds ratio and p < .05 for warmth, tenderness, erythema, induration and subcutaneous nodule. Immunogenicity IM and SC comparable response12 and no data.13
Campbell et al14	Single-center, randomized, open, phase I study.	Healthy US adults 18–40 y old. n = 80	Experimental Anthrax vaccine n = 60 Anthrax toxoid (AVA) vaccine, n = 20 IM n = 10 SC n = 10	SC > IM, p < .05 for subcutaneous nodules for AVA Immunogenicity Peak antibody, SC and IM comparable.
Pondo et al15	Multi-center, randomized, double-blind, phase IV study.	Healthy US adults 18–61 y old n = 1564	Anthrax toxoid (AVA) vaccine administered according to 7 different protocols.	Reactogenicity SC > IM, p < .05 for warmth, tenderness, erythema, induration and subcutaneous nodule. Immunogenicity IM not inferior to SC administration.
Edelman et al16	Randomized, double-blind, Phase II study.	US adults 18–40 y old n = 144	Clostridium botulinum type F toxoid vaccine. Data for 116 patients. Total number of injections n = 419 IM n = 167 SC n = 252	Reactogenicity SC > IM, p < .5 for subcutaneous nodules at primary injection. Immunogenicity Similar immune response in both SC and IM groups.
Carlsson et al17	Multi-center, randomized, open study.	Swedish infants, 3 months old. n = 287	D, DT, DT/inactivated polio (IPV) vaccine reconstituted with Haemophilus influenzae type b, Hib-T (Act-Hib) Data for: n = 365 (injections.) IM n = 184 SC n = 181	Reactogenicity SC > IM, p < .05 for pain, redness and subcutaneous nodules. Immunogenicity IM and SC comparable response
Volk et al18	Multi-center, observational study.	US children and adults. Ages not given. Adults, n = 1338. Children, n = 2126	Toxoid antigen 3 or 5 antigen preparations; 3 contained diphtheria, pertussis and scarlet fever; 5 contained the above 3 as well as tetanus and typhoid antigens. Data for injections, n = 9236 IM n = 6760 SC n = 2376	Reactogenicity SC > IM, p < .5 for sterile abscess (antigen cysts). Immunogenicity No data recorded
Mark et al19	Multi-center, randomized, open study.	Healthy Swedish infants, 3 months old. n = 252	Diphtheria/tetanus toxoid (DT) vaccine. Data for n = 243 IM n = 122 SC n = 121	Reactogenicity SC > IM, p < .5 for Redness and swelling. SC > IM for pain immediately after injection, but not statistically significant. Immunogenicity IM and SC comparable response.
Rothstein et al20	Multi-center, randomized, double-blind study.	US infants, 3 months old. n = 80	Diphtheria/tetanus/ acellular pertussis (DTaP) vaccine. Data for n = 80 IM n = 40 SC n = 40	Reactogenicity SC > IM, p < .05 for redness with 1st, 2nd and 3rd dose. Immunogenicity IM and SC comparable response.
Diggle & Deeks21	Multi-center, randomized, single-blind study.	UK infants, 4 months old. n = 119	Diphtheria/tetanus/ whole cell pertussis (DTwP) vaccine plus HibTITER vaccine. Data for 110 IM n = 53 SC n = 57	Reactogenicity SC > IM, p < .05 for redness and swelling. Immunogenicity No data recorded.
Diggle et al22	Multi-center, randomized, single-blind study.	UK infants 2, 3 and 4 months old. n = 564	DTwP/Hib administered concomitantly with meningococcal C vaccine into contralateral thigh. Data for n = 368 IM n = 189 SC n = 179	Reactogenicity Significantly less local reactions for IM compared with SC. Immunogenicity IM and SC comparable response.
Jackson et al23	Multi-center, open, non-randomized, post licensure safety study.	US children 4–6 y old. n = 1315	DTaP vaccine. Data for n = 1315 IM n = 985 SC n = 430	Reactogenicity SC > IM, p < .05 for redness, swelling and pain. Immunogenicity No data recorded
Ipp et al24	Multi-center, open, non-randomized study.	US children 18 months old. n = 205	DTwP- Polio vaccine. Route comparative study. Data for n = 131 IM n = 67 SC n = 64	Reactogenicity SC > IM, p < .05 for any reaction, any swelling. Immunogenicity No data provided.
Holt & Bousfield25	Multi-center, open, non-randomized study.	English children, age data not clearly defined. n = 895	Diphtheria toxoid vaccine (PTAP) Data for n = 895 IM n = 556 SC n = 339	Reactogenicity No data supplied. Immunogenicity IM gave significantly greater Schick conversion rate than SC injection.
Ragni et al26	Multi-center open, randomized, phase IV study.	US patients 2–18 y old with hemophilia compared with non-hemophilic siblings. n = 86	Inactivated, adjuvanted Hepatitis A (HAV) virus vaccine. Data for n = 86 IM n = 41 non- hemophilic siblings. SC n = 45, patients with hemophilia. M > F, p = ≤0.05 hemophilia patients compared with non-hemophilic siblings.	Reactogenicity SC > IM, but p > .05 for swelling. Immunogenicity IM > SC. GMT, anti-HAV At 1 month: 233mIU/ml, 185mIU/ml At 8 months: 1022mIU/ml, 584mIU/ml
Frosner et al27	Two single- center, open, randomized pilot studies. One compared IM with SC administration.	Healthy Swiss adults, 18–45 y old. n = 115	Virosomal, adjuvanted hepatitis A vaccine. Data for n = 115 IM n = 71 SC n = 44	Reactogenicity SC > IM, but p > .05 for local reaction, pain and tenderness after primary vaccination. Immunogenicity Seroconversion: IM 95.8% vs SC 93.2%.
Fisch et al28	Two-center, open, randomized study	French adults 19–59.6 y old. n = 147	Inactivated, adjuvanted Hepatitis A (HAV) vaccine Given by IM or SC by needle injection: Data for n = 99 IM n = 50 SC n = 49	Reactogenicity SC > IM, p < .05 for local reaction. Immunogenicity IM and SC comparable response.
Parent du Chatelet et al29	Multi-center, randomized study	French adults, 18–60 y old. n = 138	Inactivated, adjuvanted Hepatitis A vaccine. Given by IM or SC needle injection. Data for n = 92 IM n = 46 SC n = 46	Reactogenicity SC > IM, p < .05 for redness. Immunogenicity IM and SC comparable response.
Ogawa et al30	Retrospective study.	Healthy Japanese University students, age 19–30 y old. n = 1135	Inactivated, adjuvanted Hepatitis B vaccine. Data for n = 620 IM n = 247 SC n = 373	Reactogenicity No data supplied. Immunogenicity Significantly better seroconversion IM vs SC At 2 months: IM 84.6%, SC 62.7% At 5 months: IM 93.5%, SC 77.0%
Probst et al31	Single center, randomized study.	Swiss hemodialysis adult patients, aged 47–50 ± 14 y old. n = 81	Adjuvanted, recombinant Hepatitis B vaccine. Data for n = 54 IM n = 27 Deltoid muscle. SC n = 27 Volar aspect of forearm.	Reactogenicity No data supplied. Immunogenicity Seroconversion: IM 76%, SC 69% GMT, HBsAb IM 443 mIU/ml SC 79 mIU/ml
Yamamoto et al32	Single center, open, randomized, phase I study.	Healthy Japanese adults ≥ 18 y old. n = 124	Adjuvanted, recombinant Hepatitis B vaccine. Data for n = 124 IM n = 62 SC n = 62	Reactogenicity No data supplied. Immunogenicity Seroconversion: IM 98%, SC 97% GMT, HBsAb, IM > SC, IM 791mIU/ml SC 168mIU/ml
Suzuki et al33	Single center, phase I, multicenter, phase II and III, open, non- randomized studies.	Japanese patients, children ≥ 10 y old and adults. n = 2137	Yeast derived, adjuvanted, recombinant, pre S and S containing Hepatitis B vaccine. Data for injections n = 4723 IM n = 2693 SC n = 2030	Reactogenicity SC > IM, p < .05 for pain, redness, swelling and warmth. Immunogenicity At 7 months: IM > SC GMT, HBsAb: IM 1396mIU/ml SC 748mIU/ml Anti-pre S2: IM1185mIU/ml SC 566mIU/ml
Kishino et al34	Multicenter, randomized study.	Healthy Japanese adults. Age 20–35 y old. n = 383	Recombinant, inactivated adjuvanted Hepatitis B vaccine. Data for n = 383 IM n = 94 SC n = 279	Reactogenicity SC > IM, p < .5 for pain, redness, swelling and pruritis. Immunogenicity Seroconversion: IM 98.7%, SC 91.6% GMT, HBsAb IM 1064mIU/ml SC 231.5mIU/ml
De Lalla et al35	Single center, open, randomized study.	Healthy Italian adults, age range 26.3–28 y old. n = 151	Adjuvanted, recombinant Hepatitis B vaccine. Data for n = 151 IM n = 75, SC n = 76	Reactogenicity SC > IM, p > .05 Immunogenicity Seroconversion: IM 88% SC 75%
Carpenter et al36	Retrospective study.	US children with bleeding disorders, n = 207 Testing for HbsAb was done at: SC 53 ± 20 months, IM 60 ± 20 months after vaccinations, p = .02 for time after vaccination.	Adjuvanted Hepatitis B vaccine Data for n = 206 IM n = 114 SC n = 92	Reactogenicity SC > IM, p > .05 for intramuscular hematoma Immunogenicity IM and SC comparable response.
Vink et al37	Single center, open-label, randomized, Phase III study.	Japanese adults, mean age 61.9 y old. n = 60,	Herpes zoster recombinant, adjuvanted, subunit vaccine (HZ/su) containing VZV. Data for n = 58 IM n = 29 SC n = 29	Reactogenicity SC > IM, p < .05 for redness and swelling. Immunogenicity Seroconversion rates: IM and SC 100% Anti-gE antibody Geometric mean concentration: IM 45521mIU/ml SC 44126mIU/ml
Ikeno et al38	Single center, randomized, phase I study.	Japanese males, 20–40 y old. n = 120	Inactivated, adjuvanted, monovalent, whole viruses A/H5N1 influenza vaccine. Data for n = 120 3 different doses: (1.7 µg, 5 µg, 15 µg) IM n = 20 each dose, SC n = 20 each dose. 2 doses 21 d apart.	Reactogenicity SC > IM, p < .05 for redness and swelling in 1st and 2nd dose. Immunogenicity Seroconversion: After 1st dose: 1.7 µg IM 10%, SC 0%, 5 µg IM 35%, SC 10% 15 µg IM 65%, SC 42% After 2nd dose: 1.7 µg IM 20%, SC 20%, 5 µg IM 50%, SC 20% 15 µg IM 75%, SC 68%
Hopf et al39	Single center, open, randomized, study.	Healthy Austrian adults, 18–60 y old. n = 116	Adjuvanted, Inactivated tick-borne encephalitis (TBE) virus vaccine. Data for 116 IM n = 58, SC n = 58	Reactogenicity SC > IM, p < .05 for pain, redness and swelling. Immunogenicity IM and SC comparable response

Seroconversion hepatitis B vaccine – HbsAb ≥ 10mIU/ml

Seroconversion hepatitis A vaccine – anti-HAV level ≥ 20mIU/ml

Seroconversion influenza vaccine – percentage with >4 fold increase in post-vaccination hemagglutinin inhibition (HI) titer.

Table 2.

Live virus vaccines and intramuscular compared with subcutaneous administration – reactogenicity and immunogenicity

Author	Study design	Patients	Intervention	Outcome
Bernstein et al40	Single-center, double-blind, randomized, placebo controlled phase I study.	Healthy US adults, 18–45 y old n = 40	Cytomegalovirus vaccine. IM n = 16 SC n = 16 Placebo n = 8 Low dose n = 16, IM n = 8 SC n = 8 High dose n = 16 IM n = 8 SC n = 8	Reactogenicity Redness and swelling only seen in those who received active vaccine by SC administration. Immunogenicity Similar antibody response IM and SC groups.
Diez-Domingo et al41	Multi-center, randomized, open-label study.	Healthy German and Spanish adults ≥ 50 y old n = 354	Live attenuated herpes zoster vaccine. Data for 352 IM n = 175 SC n = 177	Reactogenicity SC > IM, p < .05 for injection site reaction (0–21 d) Immunogenicity Similar antibody titers IM and SC groups.
Koblin42	Multi-center, randomized, open-label study.	US and Peru adults 18–50 y old n = 90	HIV DNA prime and booster with rAd5 vaccine. Data for n = 40 IM n = 20 SC n = 20	Reactogenicity SC > IM, p < .05 for redness/induration and pain. Immunogenicity Similar antibody titers in both IM and SC groups
Peters et al43	Double-blind, randomized, placebo-controlled, dose-escalation study.	UK and Kenya Adults. 18–59 y old n = 70 Nairobi, n = 45 London. n = 115	PTHr HIVA DNA and recombinant MVA HIVA vaccines. Data for n = 68 IM n = 35 SC n = 33	Reactogenicity SC > IM, p < .05 for Moderate/severe local reactions Immunogenicity Similar antibody titers IM and SC groups.
Enama et al44	Single-center randomized, open, phase I study.	US adults 18–50 y old. n = 60	HIV, DNA and comparator rAd5 HIV vaccines Data for DNA primes, IM n = 10 SC n = 10 Data for rAd5 prime, IM n = 10 SC n = 10	Reactogenicity HIV DNA SC > IM, p < .05 for swelling. rAd5 SC > IM, p < .05 for swelling, redness and injection site reaction, subcutaneous nodules. (SC 2, IM 0). Immunogenicity Similar antibody titers IM and SC groups.
Lafeber et al45	Single-center randomized, open study.	Dutch children 14 months old n = 67	MMR vaccine Data for n = 67 IM n = 33 SC n = 34	Reactogenicity SC > IM, p > .05 for pain immediately after injection. Immunogenicity Response to vaccine antigens not significantly different.
Kuter et al46	Post-licensure analysis of 33 studies.	Infants/children 11–18 months old. n = 752	MMRII – rHA vaccine and Varivax® (Varicella vaccine). No data for numbers given by IM or SC administration.	Reactogenicity SC > IM, p < .05 for injection site reactions. Immunogenicity Seropositivity rates after IM and SC administration were comparable.
Knuf et al47	Multi-center, randomized study.	German infants/children 11–21 months old. n = 328	MMR vaccine. Data for n = 318 IM n = 161 SC n = 157	Reactogenicity SC > IM, p > .05 Immunogenicity IM and SC comparable antibody responses for all antigens.
Gillet et al48	Multi-center, randomized, Open-label study.	French infants/children 12–18 months old. n = 752	Measles, mumps, rubella vaccine. Data for n = 712 IM n = 349 SC n = 363	Reactogenicity MMR SC > IM, p < .05 for any injection site reaction and redness. Varicella SC > IM, p < .05 for any injection site reaction and redness. Immunogenicity comparable immune response SC and IM.
Haas et al49	Multi-center randomized, open-label, Phase III study.	Healthy French infants/children 12–18 months old. n = 405	Measles/Mumps/ Rubella/Varicella vaccine. Data for n = 405 IM n = 202 SC n = 203	Reactogenicity SC > IM, p < .05 for injection site reaction. Immunogenicity GMTs comparable for SC and IM groups.
Pittman et al50	Single-center, randomized, open-label, phase I study.	Healthy US adults, at least 18 y old. n = 43	Rift Valley Fever vaccine (MP-12) IM n = 6 (103.4 pfu) SC n = 10(104.7pfu) IM n = 27(104.4pfu)	Reactogenicity SC and IM comparable for tenderness. Immunogenicity IM and SC comparable
Wilck et al51	Multi-center, randomized, open, dose escalation study.	US adults 18–34 y old. n = 72	Live attenuated Vaccinia vaccine; Modified Vaccine Ankara (MVA) 107 or 108 TCID50 Data for n = 40 IM n = 20 SC n = 20	Reactogenicity SC > IM, p < .05 for severe local erythema and induration (31–70 mm) Immunogenicity IM and SC comparable antibody responses.
Vollmar et al52	Single-center, randomized, double-blind, phase I study.	Healthy German males 20–55 y old n = 86	Live attenuated Vaccinia vaccine MVA-BN 108 TCID50 Data for n = 36 IM n = 18 SC n = 18	Reactogenicity SC > IM, p < .05 for Redness and swelling Immunogenicity IM and SC comparable antibody responses.
Frey et al53	Single-center, randomized, partially-blinded, phase I study.	Healthy US adults aged 18–32 y old n = 90	Live attenuated Vaccinia vaccine MVA-BN 108 TCID50 Data for n = 30 IM n = 15 SC n = 15 for each group.	Reactogenicity 1st dose SC > IM, p < .05 for redness and induration Immunogenicity IM and SC comparable antibody responses.
Seaman et al54	Single-center, randomized, double-blind, placebo controlled study.	Healthy US adults 18–34 y old. n = 36	Live attenuated MVA Vaccinia vaccine. 107 TCID50 challenge with Vaccinia vaccine Dryvax® Data for n = 12 IM n = 5 SC n = 7	Reactogenicity No data supplied Immunogenicity SC < IM, p > .05
Dennehy et al55	Two-center, randomized, study.	US infants and children, 12 months – 10 y old n = 132	Varicella vaccine. Data for n = 132 IM n = 67 SC n = 65	Reactogenicity SC > IM, p < .05 for injection site reaction Immunogenicity GMTs comparable for IM and SC administration.
Fox et al56	Non- randomized study.	Brazilian male military personnel. 15–40 y old. n = 552	Yellow Fever Vaccine 17D-NY104, dose escalation, route comparative studies. Minimum immunizing dose assessed as a 50% lethal dose of a mouse lot.	Reactogenicity No data supplied Immunogenicity Minimum Immunizing Dose (mid) IM 1.6 SC 2.5

pfu – plaque forming units

TCID_{50 –} Median tissue culture infectious dose

mid – minimum immunizing dose

Table 3.

Non-adjuvanted (whole cell, split cell and subunit) vaccines and intramuscular compared with subcutaneous administration – reactogenicity and immunogenicity

Author	Study design	Patients	Intervention	Outcome
Leung et al57	Non-randomized study. Every 2nd child given SC injection.	Canadian children 15 months to 5 y. n = 498	Inactivated, whole cell Haemophilus influenzae type b polysaccharide vaccine. Data for n = 398 IM n = 194 SC n = 194	Reactogenicity IM > SC, p < .05 for crying Immunogenicity No data supplied
Cook et al58	Single-center, randomized, observer-blind study.	Australian adults ≥65 y old, 55 y old if had physician diagnosed chronic disease. n = 720	Split-virus influenza vaccine. Data for n = 709 IM n = 356 SC n = 353	Reactogenicity SC > IM, p < .05 for redness, swelling and tenderness. Immunogenicity Seroconversion: H3N2 IM 80.5%, SC 71.1%, p = .0045. H1N1 IM 37.2%, SC 26.9%, p = .0043. B, IM 57.0%, SC 51.0%, p = .1948.
Ruben & Jackson59	Multi-center, randomized study.	US Adults 18–25 y old with small number of older subjects.	Four subunit influenza vaccines, A2/Aichi and B/Mass. No number given for IM and SC injection.	Reactogenicity SC > IM – 2 fold for local pain. SC > IM – 8 fold for erythema and induration. Immunogenicity Fold increase in titer (post:pre vaccination) A2/Aichi: IM 20.5, SC 6.8 A2/Aichi vs B/Mass IM 20.5 and 8.0 respectively.
Sanchez et al60	Two-center, randomized, phase I/II, double-blind study.	Japanese adults ≥65 y old. n = 120	High dose, split virus influenza vaccine. Data for n = 110 IM n = 55 SC n = 55	Reactogenicity SC > IM for injection site pain, erythema, swelling and induration. p < .05 Immunogenicity Fold increase: H3N2 IM 16.93, SC8.31 H1N1 IM 16.0, SC 9.25 B Yamagata IM 7.51, SC 4.68 B Victoria IM 10.69, SC 6.92
Delafuente et al61	Multi-center, randomized, single-blind study	Elderly males, mean age 68 y old, range 61–81 y. On warfarin anticoagulant. n = 26	Split virus influenza vaccine, 1991–1992. Data for n = 26 IM n = 13, SC n = 13	Reactogenicity No difference in adverse events between IM and SC administration. Immunogenicity Comparable immune response in IM and SC.
Ballester-Torrens et al62	Single-center, randomized, single-blind, phase IV study	adults n = 59	Split-virus influenza vaccine. Data for n = 59 IM n = 30 SC n = 29	Reactogenicity SC > IM, p < .05 for local reaction and pain. Immunogenicity No data provided
Casajuana et al63	Multi-center, randomized, single-blind study.	Spanish adults older than 18 y on oral anticoagulants n = 229	Split virus influenza vaccine. Data for n = 207 IM n = 92 SC n = 115	Reactogenicity SC > IM, p < .05 for erythema Immunogenicity No data supplied
Laurichesse et al64	Single-center, double-blind, randomized, placebo-controlled study.	French adults 18–40 y old. n = 84	Inactivated, whole cell Leptospira interrogans (Serogroup icterohaemorrhag-iae) vaccine. Data for n = 60 IM n = 30 SC n = 30	Reactogenicity SC > IM, p < .05 for local reaction at 14 d. Immunogenicity Similar antibody response for IM and SC routes.
Ruben et al65	Single-center, randomized study.	US adults. Mean age: IM 21.9 y old SC 20.6 y old n = 141	Inactivated, whole cell meningococcal vaccine (A,C,Y,W-135). Data for n = 132 IM n = 66 SC n = 66	Reactogenicity SC > IM, p < .05 for erythema. Immunogenicity IM and SC comparable response.
Scheifele et al66	Single-center, non-randomized study.	Canadian children 4–6 y old n = 101	Inactivated, whole cell meningococcal polysaccharide vaccine (A, C, Y, W-135) First 53 given SC immunization, subsequent 48 given IM immunization.	Reactogenicity SC > IM, p < .05 for any redness or swelling. Immunogenicity No data provided
Cook et al67	Single-blind, randomized, prospective trial	Australian adults ≥65 y old, 55 y old if had physician diagnosed chronic disease. n = 254	Inactivated, whole cell pneumococcal 23 valent vaccine Data for n = 254. IM n = 127 SC n = 127	Reactogenicity SC > IM, odds ratio 3.2 95% CI (1.13–1.93) Immunogenicity Comparable antibody response IM and SC route.

Seroconversion influenza vaccine – percentage with > 4 fold increase in post-vaccination hemagglutinin inhibition(HI) titer.

Fold increase influenza vaccine – Ratio of post- to pre-vaccination titer.

Thirty studies^10–39 comparing intramuscular with subcutaneous administration of adjuvanted vaccines are presented in alphabetical order in Table 1 (6 anthrax^10–15, 1 botulinum toxoid,¹⁶ 9 diphtheria and tetanus toxoid containing vaccines,^17–25 4 hepatitis,^26–29 7 hepatitis, ^30–36 1 herpes zoster,³⁷ 1 influenza³⁸ and 1 tick-borne encephalitis³⁹). These studies could be subdivided into two groups; one with 21 randomized trials and the other with 7 observational studies and 2 randomized trials with unacceptable biases.

The 21 randomized trials being; 6 anthrax¹⁰⁻¹⁵, 1 botulinum toxoid,¹⁶ 5 diphtheria toxoid containing vaccines,^17,19-22 3 hepatitis, ^27–29 3 hepatitis, ^32,34,35 with 1 each of herpes zoster,³⁷ influenza³⁸ and tick-borne encephalitis³⁹ vaccines. There were 7 observational studies. These were 4 diphtheria/tetanus toxoid containing vaccines^18,23-25 and 3 hepatitis B vaccines.^30,33,36

Two studies were excluded from the randomized trial group due to unacceptable biases. In the study, Ragni et al.²⁶ with hepatitis A vaccine, patients with hemophilia were given SC injection and compared with non-hemophilic siblings given IM injection. Whilst in the study by Probst et al.³¹ with hepatitis B vaccine IM injection was given into the deltoid muscle and SC injection was given into the volar surface of the forearm.

Five studies^20–24 with diphtheria/tetanus toxoid were included where the vaccines were given with a 16 mm compared with 25 mm long needle as the former was considered to give SC injection and the latter to give IM injection.

In the 21 randomized trials, local reactogenicity data were provided in 20 studies. In 18 studies,^{10–17,19–22,28,29,34,37-39} SC injection gave significantly greater rates of reaction than IM injection. In two other studies,^27,35 SC gave greater rates of reaction than IM injection but this did not reach statistical significance. Subcutaneous nodules were significantly more frequent for SC compared with IM injection for anthrax vaccine¹⁰⁻¹⁵, botulinum toxoid vaccine¹⁶ and a combination diphtheria toxoid vaccine.¹⁷ In an observational study¹⁸ with diphtheria toxoid containing vaccines, sterile abscess formation was significantly greater for SC compared with IM injection.

Pain immediately after injection (assessed with a pain analogue scale) was reported¹¹ to be significantly less for a 4 IM regimen of anthrax vaccine compared with a 4 SC regimen. Mark et al.¹⁹ reported a similar trend but this did not reach statistical significance.

Immunogenicity data were recorded in 19 of the randomized trials.^{10–12,14–17,19,20,22,27-29,32,34,35,37-39} Immunogenicity was greater for IM compared with SC injection in six studies^{27,32,34,35,37,38} being significantly greater in the studies by Kishino et al.³⁴ (hepatitis B vaccine) and Ikeno et al.³⁸ (first dose of an influenza vaccine). In the remaining 13 studies^{10–12,14–17,19,20,22,28,29,39}, the immune response was comparable for IM and SC injection.

Seventeen studies comparing IM with SC administration of live virus vaccines are presented in alphabetical order in Table 2 (1 cytomegalovirus,⁴⁰ 1 herpes zoster,⁴¹ 3 human Immunodeficiency virus,^42–44 5 measles-mumps-rubella,^45–49 1 Rift Valley fever,⁵⁰ 4 vaccinia,^51–54 1 varicella⁵⁵ and 1 yellow fever⁵⁶). Fifteen of the 17 studies were randomized trials.^{40–45,47-55}

In 13 studies^{40-44,47–53,55} out of the15 studies where reactogenicity data were provided, SC injection gave significantly greater rates of local reaction than IM injection. In the study by Lafeber et al.,⁴⁵ pain immediately after injection was greater with SC compared with IM injection but this did not reach statistical significance. Two subcutaneous nodules were observed following SC injection of one HIV vaccine⁴⁴ but not with IM injection.

IM and SC immunogenicity data were comparable in 15 randomized trials.^{40–45,47-55} Immunogenicity was greater for IM compared with SC injection in one study.⁵⁴ In this study by Seaman et al.⁵⁴ immunogenicity was greater for IM compared with SC injection but this did not reach statistical significance.

Eleven studies comparing IM with SC administration of non-adjuvanted, inactivated (whole cell, split cell and subunit) vaccines are presented in alphabetical order in Table 3 (1 Hemophilus influenzae type b,⁵⁷ 6 influenza,^58–63 1 leptospirosis,⁶⁴ 2 meningococcal,^65,66 1 pneumococcal⁶⁷).

Nine of the 11 studies were randomized trials.^58–65,67 In 8^{58–60,62-65,67} of the 9 studies where reactogenicity data were provided, SC injection was associated with significantly greater rates of reaction than IM injection. In seven of the nine randomized trials where immunogenicity data were provided,^{58–61,64,65,67} IM gave comparable results with SC injection in four studies.^61,64,65,67 In three studies,^58–60 antibody response was significantly greater for IM compared with SC injection for influenza A.

Discussion

This extensive narrative review provided high-grade evidence⁹ that intramuscular (IM) injection significantly reduced the likelihood of local reactogenicity compared with subcutaneous (SC) injection. High-grade evidence was drawn from studies with all vaccine types (adjuvanted n = 18, live virus n = 13, non-adjuvanted inactivated (whole cell, split and subunit) n = 8).

The greater rates of reactogenicity were also seen for vaccines recommended⁵ to be given by SC injection (quadrivalent meningococcal polysaccharide (4vMenPV), varicella (VV), measles-mumps-rubella/varicella (MMR/V), herpes zoster vaccine) and vaccines recommended to be given by either IM or SC route (influenza and 23-valent pneumococcal (23vPPV)).

Direct route comparative studies have not been reported for inactivated polio (IPV), Japanese encephalitis (Imojev®), Q fever and rabies vaccine. Studies with IPV^17,68 given IM or SC with other antigens have shown comparable immunogenicity for IPV. Consequently, the recommendation for IPV alone to be given by SC injection is inconsistent with these data.

Older rabies vaccines were derived from animal neural tissue and given by subcutaneous injection.⁶⁹ Currently recommended⁷⁰ rabies vaccines are derived from cell cultures and are given by IM injection. The latter are more immunogenic and associated with less severe adverse reactions than the older rabies vaccines.

Subcutaneous nodules are uncommonly reported in this review and almost entirely with adjuvanted vaccines (anthrax¹⁰⁻¹⁵, botulinum toxoid¹⁶ and diphtheria combination vaccine¹⁷). A single report⁴⁴ of the transient formation of two nodules was made with an HIV vaccine. Subcutaneous nodules have been considered⁷¹ to be benign, self-limiting AEFIs but this is clearly not the case as demonstrated by Bernstein et al.⁷² who reported 11.4% of nodules persisting at 180 d post anthrax vaccination. These nodules may persist⁷³ for years and are often associated with pruritis and superficial dermatological features such as eczema, lichenification and hyperpigmentation.

Route of administration and use of aluminum salt adjuvants are recognized⁷¹ determinants of their formation. However, the role of aluminum hydroxide sensitivity in the pathogenesis of these nodules is controversial with some authors demonstrating this phenomenon⁷⁴ whilst others⁷⁵ claiming that nodule formation reflects SC rather than IM injection of aluminum adjuvanted vaccines. Sterile abscess formation was also significantly greater with SC than IM injection for an adjuvanted diphtheria toxoid vaccine in an observational study.¹⁸

Pain immediately after injection might be expected⁷⁶ to be greater with IM compared with SC injection as the former has a dense supply of nociceptive nerve endings with the subcutaneous space being relatively devoid of pain receptors. Pain assessed (using standardized pain assessment scales) was significantly greater with SC than IM with anthrax vaccine¹¹ in this review. The same trend was seen with MMR⁴⁵ and DT toxoid vaccines¹⁹ using the same methodology but did not reach statistical significance.

This review provided moderate grade evidence that IM injection significantly improved the immunogenicity of vaccines compared with SC injection. This grade of evidence was drawn from better antibody response/seroconversion data with adjuvanted vaccines n = 6, live virus vaccines n = 1 and non-adjuvanted, inactivated (whole cell, split and subunit) vaccines n = 3 for IM compared with SC injection. In this review, no study with SC injection was observed to be more immunogenic than IM injection. The extent and availability of the immunogenicity data were influenced by trial design factors (e.g. set to demonstrate non-inferiority between routes of administration and Phase I studies)

Phase I studies⁷⁷ are safety and tolerance studies with one of their objectives to identify preferred routes of administration. In the randomized trials of this review, 33% had less than 100 patients (3/21 adjuvanted vaccines,^14,20,37 9/15 live virus vaccines^{40,42,44,45,50-54} and 3/9 non-adjuvanted, inactivated (whole cell, split cell and subunit vaccines)).^61,62,64

The combination of high-grade reactogenicity evidence with the moderate grade immunogenicity evidence allows a strong recommendation⁷⁸ that all vaccines, except BCG (intradermal) and rotavirus (oral), should be given by IM injection. This will simplify vaccination practice and prevent the inadvertent misadministration of vaccines (meningococcal conjugate vaccine⁷⁹ and recombinant zoster vaccine⁸⁰). It may potentially reduce vaccine hesitancy/refusal^2,3 due to a lower rate of ISRs with IM compared with SC injection.

The use of evidence-based medicine in vaccinology should replace highly idiosyncratic and divergent practices that are outmoded by promoting accountability based on best scientific principles.