Background
Timely and complete vaccination with multi-dose schedules is of public health importance, because an incomplete vaccination series may yield suboptimal disease protection. Most studies investigating completion of multi-dose vaccine schedules have been conducted for pediatric and adolescent populations [
1‐
3], for whom several different multi-dose vaccines are recommended. Instead little information is available regarding adherence to and completion of multi-dose vaccine series among adults.
Hepatitis A and hepatitis B are vaccine-preventable viral diseases of the liver [
4‐
7] for which effective multi-dose vaccines have been available for many years. In the United Kingdom (UK), hepatitis A and/or hepatitis B vaccinations are recommended for adults at high risk of exposure to the specific virus or of complications from the disease [
8,
9], among them travelers to high-risk countries, health care providers, intravenous drug users, sex workers, prisoners, patients with chronic liver disease, and any persons at risk of contact with blood or body fluids [
10]. All recommended vaccinations are provided for free by the UK National Health Service (NHS). The usual recommended vaccination schedules are for two doses of the hepatitis A vaccine within 12 months, and for three doses of the hepatitis B or combination hepatitis A/B vaccine within 6 months.
Studies of adults in countries of low hepatitis prevalence, such as the UK, United States (US), Canada, and western Europe, indicate that hepatitis A and B vaccine series completion rates among adults are low [
11‐
16]. For example, in a large 2015 survey in the US, only 16 and 32% of travelers to countries of high to intermediate endemicity had completed the two- or three-dose series for hepatitis A or hepatitis B, respectively [
11]. The majority of studies of hepatitis vaccine series completion, however, are done in circumscribed populations such as the homeless [
14], travelers [
15], and adults with chronic liver disease [
16].
There are few studies of series completion for multi-dose vaccinations in broad, general adult populations, and few studies have examined the rates of vaccine series completion within the recommended time frames [
13], including none in the UK. We conducted a population-level analysis of vaccination completion rates using a large routinely collected dataset in the UK [
17]. The primary objective of this study was to estimate dose and series completion for hepatitis A, hepatitis B, and combination hepatitis A/B multi-dose vaccines and adherence to the two- and three-dose vaccination schedules among adults in the UK receiving care under the umbrella of the NHS.
Methods
Data source
This retrospective cohort study used anonymized electronic health record (EHR) data from the Clinical Practice Research Datalink (CPRD), a large, well-managed database maintained by the UK Department of Health and Social Care [
17]. Used frequently for pharmacoepidemiological research, the CPRD contains longitudinal EHR data that originates from over 600 subscribing general practices throughout the UK, representing about 5 million patients with active EHRs, or 7% of the UK population.
We used CPRD data from January 1, 2009, through December 31, 2016. The protocol for this study was approved by the CPRD Independent Scientific Advisory Committee (ISAC reference number 17_226R). No patient identifying information was accessible during the study.
Study population
Adults who had a recorded dose of hepatitis A, hepatitis B, or hepatitis A/B vaccine at 19 years of age or older were eligible for the study. We required individuals receiving the hepatitis A vaccine to have at least 12 months of continuous baseline data before their first identified dose, while those receiving hepatitis B or hepatitis A/B vaccine were required to have at least 6 baseline months. The 12-month (hepatitis A) and 6-month (hepatitis B and A/B) baseline data requirements were selected to maximize the possibility that we identified the initiation of a vaccination series in the database rather than a booster.
Patients on hemodialysis who received an altered hepatitis B vaccination schedule or dosage were excluded from the study. In addition, individuals who received a second dose of the hepatitis A/B vaccine within 2 weeks of the first dose were excluded, as this schedule indicates use of accelerated administration requiring a total of four vaccines.
Outcome measures
The primary study outcomes were (1) completion of the recommended number of doses of the relevant hepatitis vaccine (series completion) and (2) adherence to the recommended timing of the 2-dose hepatitis A and 3-dose hepatitis B and hepatitis A/B vaccines as per UK labeling for each vaccine product (Table
1) [
18]. We calculated the proportions of adults who completed two and three doses and the proportions who adhered to the recommended schedule or within prespecified additional time periods after the first dose, as outlined in Table
1. In addition, we estimated the median time to completion of each dose and each vaccination series.
Table 1
Recommended adult administration schedules for hepatitis A, B, and A/B vaccine products in the United Kingdom
Hepatitis A | 0 | 6–12 mob | n/a |
Dose 2:
At 24 mo and 36 mo |
Hepatitis B | 0 | 1 mo | 2–6 mo |
Dose 2:
At 13 mo and 25 mo |
Hepatitis A/B | 0 | 1 mo | 6 mo |
Dose 3:
At 18 mo and 30 mo |
Statistical analysis
The study population was stratified by first vaccine type (hepatitis A, B, or combination A/B), and each vaccine cohort was analyzed separately. Therefore, if an individual received both hepatitis A and hepatitis B vaccines during the study period, that individual would be included in both vaccine cohorts. Results for series completion and adherence were evaluated for the full cohorts and stratified by age at first dose within each cohort (19–49, 50–59, 60–64, 65–69, and ≥ 70 years). Baseline and outcome measures were summarized descriptively using means, medians, 95% confidence intervals (CIs), standard deviations, and frequency distributions for categorical variables.
We used Kaplan-Meier (KM) methods to estimate the time to completion of the second and third dose (as appropriate) at prespecified time points after series initiation. Only individuals with complete information for all the variables in the model were included in the study; hence there was no imputation for missing data. The use of the KM method allowed for inclusion of full vaccine cohorts by right-censoring of records for those without evidence of series completion. These individuals were censored at the end of the data. Survival function by age group was compared with the reference group of oldest adults (age ≥ 70 years) by calculation of hazard ratios (HR) with 95% CI.
We conducted sensitivity analyses to assess the impact of requiring a longer continuous baseline (pre-vaccine) period of 24 months’ data for the hepatitis A cohort and 18 months’ data for hepatitis B and hepatitis A/B cohorts. In addition, we analyzed series completion for individuals in each vaccine cohort with continuous follow-up data after the first vaccine dose of 6, 12, 18, and 24 months.
Discussion
The rates of adherence to and series completion of UK vaccination schedules for multi-dose hepatitis A and hepatitis B vaccines were low among adults in this retrospective observational study. The percentages of adults adhering to and completing the recommended multi-dose series were 11, 22, and 10% for hepatitis A, hepatitis B, and hepatitis A/B vaccines, respectively. These percentages rose to only 23, 35, and 33%, respectively, when the follow-up periods were extended to 36 months for hepatitis A and to 30 months for hepatitis B and A/B vaccines, well beyond the recommended windows for vaccine series completion. Sensitivity analyses requiring longer periods of continuous data before and after vaccine series initiation supported the primary findings, with adherence rates within 2 percentage points. Overall, these analyses suggest substantial waste of health care resources, with vaccines delivered but likely suboptimal protection obtained because of incomplete vaccination series.
We found that individuals in the older age groups (≥50 years old) tended to be more adherent than younger adults (19–49 years old). This may be because older adults received the hepatitis vaccine together with an influenza vaccine, recommended annually in the UK for individuals 65 years and older, or together with the shingles vaccine, recommended for adults at age 70 [
10]. Both influenza and shingles vaccine are administered at no cost under UK NHS care [
19]. The hepatitis A vaccine is recommended most for travelers, while hepatitis B is recommended for both travelers and health-care providers, which could explain the considerably better adherence rates for the second dose of the hepatitis B vaccine (46%) and combination hepatitis A/B vaccine (56%), as compared with the second dose of hepatitis A vaccine (11%).
Few studies have examined multi-dose vaccine series completion in general adult populations. The recent study of Trantham et al. [
13], designed similarly to the present study but using administrative claims data, reported somewhat higher, but still poor, rates of adherence in the US than in the UK, namely, rates of adherence to recommended schedules of 27, 30, and 18% for hepatitis A, hepatitis B, and hepatitis A/B, respectively. In addition, similar to the present study, they found that older adults had better adherence rates. The earlier study of Nelson et al. [
20] in the US also found better adherence among older adult age groups and, for those adherent in all age groups, relatively long intervals between vaccine doses as in the present study and others [
13].
Prior studies have reported suboptimal rates of series completion for hepatitis vaccine also in defined subpopulations, including patients with chronic liver disease [
16], former prisoners [
14], intravenous drug users [
21], the homeless [
22], and travelers [
15,
23,
24]. Most of these studies did not examine series completion with regard to the recommended timelines.
Identification of the factors associated with adherence and dose/series completion was outside the scope of the present study; however, initiatives to improve patient and health care provider reminder systems, as well as opportunistic vaccination at other appointments, could be helpful to increase adherence and vaccination series completion. In a prior small study, Reynolds et al. [
25] investigated factors associated with poor completion rates in a low-income minority population. They found that receipt of fewer doses of vaccine was associated with being male and having severe negative emotions, among other factors. The rate of three-dose series completions was only 31% despite the vaccinations being offered at no cost. Similarly, in our study, cost would not have been a potential barrier, because the NHS provides the vaccines for free. Offering financial incentives to encourage the return for second and third doses improved adherence in two studies of intravenous drug users [
26,
27]; however, financial incentives were determined to not be a factor in series completion in a study of former prisoners [
14]. Dedicated clinics have reported improved adherence when using multiple interventions, including SMS text reminders [
28]. In addition, health-care provider recommendations to vaccinate are thought to be important, although studies find that providers, even at academic centers, often fail to adhere to hepatitis vaccination guidelines for patients with chronic liver disease [
29,
30].
This study has several strengths. We used a large, high-quality database considered to have reliable capture of administered vaccinations [
31] and that includes EHRs for a geographically diverse, representative UK population [
32]. We studied almost half a million adults initiating hepatitis vaccination series, excluding patients receiving hepatitis B vaccination while on hemodialysis as they may not be representative of the general adult population. In addition, we used the KM method, used in few prior studies [
13,
33,
34], to determine time to completion of second and third doses. The study design enabled us to identify the probable first vaccine dose by requiring baseline data periods of 6 to 12 months, extended to 18 to 24 months, depending on vaccine type, in sensitivity analyses. The long follow-up, including sensitivity analyses examining continuous data for up to 36 months after the first vaccine, enabled us to study series completion as well as to examine in detail whether these multi-dose vaccines were administered at the recommended intervals. Of course, patients could receive a booster vaccine even 5 years after the original course, although the results of our 36-month follow-up analyses suggest that overall findings would not change. However, we acknowledge that only 55–58% of patients had available data for ≥36 months after vaccine initiation.
A limitation of using the CPRD is that the EHR data are recorded for clinical purposes rather than specifically for research purposes. Moreover, as the CPRD data comprise EHRs from general practices, only the vaccinations received at those practices would be reliably recorded, while vaccinations administered in other settings such as hospitals, pharmacies, travel clinics, or occupational health services may have been missed. As for all observational studies, there is the potential for selection bias and possibility of unmeasured confounders. Another limitation is related to the comparison of completion for the selected two-dose and three-dose vaccines. The hepatitis A series has a longer window for completion than the combined A/B series, which could lead to higher completion within the recommended window. Finally, we did not examine, hence do not report, characteristics of the study population beyond sex, age, and UK region.
These limitations raise questions that would benefit from further study. In particular, a more complete capture to include all vaccinees from settings other than general practices and the reasons for these individuals initiating hepatitis vaccination series would be of great interest. We were unable to locate published population-level data for the UK regarding other settings, such as addiction or homeless services, and the reasons for hepatitis vaccinations (eg, for high-risk individuals, health care workers, travelers, etc.). As a corollary, it would be of interest to understand geographical vaccination patterns (eg, urban vs countryside) to identify areas of need. Finally, as noted above, better understanding of factors affecting adherence and of effective initiatives to improve adherence are needed to counter the potential waste of health care resources that occurs with suboptimal protection from incomplete vaccination series.