Elsevier

Vaccine

Volume 25, Issue 25, 15 June 2007, Pages 4775-4783
Vaccine

Validity of self-reported influenza and pneumococcal vaccination status among a cohort of hospitalized elderly inpatients

https://doi.org/10.1016/j.vaccine.2007.04.015Get rights and content

Abstract

Use of self-reported vaccination status is commonplace in assessing vaccination coverage for public health programs and individuals, yet limited validity data exist. We compared self-report with provider records for pneumococcal (23vPPV) and influenza vaccine for 4887 subjects aged ≥65 years from two Australian hospitals. Self-reported influenza vaccination status had high sensitivity (98%), positive predictive value (PPV) (88%) and negative predictive value (NPV) (91%), but low specificity (56%). Self-reported 23vPPV (previous 5 years) had a sensitivity of 84%, specificity 77%, PPV 85% and NPV 76%. Clinicians can be reasonably confident of self-reported influenza vaccine status, and for positive self-report for 23vPPV in this setting. For program evaluation, self-reported influenza vaccination coverage among inpatients overestimates true coverage by about 10% versus 1% for 23vPPV. Self-report remains imperfect and whole-of-life immunisation registers a preferable goal.

Introduction

Influenza vaccine and 23-valent pneumococcal vaccine (23vPPV) have been recommended for elderly persons in a number of countries for some time [1], [2]. In Australia, these vaccines have been funded nationally since 1998 and 2005, respectively [3], while in the state of Victoria, 23vPPV has been funded since 1998 [4]. Evaluation of the impact of these programs on coverage requires accurate measurement of vaccination status. Self-reported vaccination status is frequently used in public health program evaluation, as well as in clinical research and clinical practice as a marker of true vaccination status. There are in fact few practical alternatives where no register of adult vaccinations exists. If self-reported status 23vPPV and influenza vaccination status are to be used to guide practice their validity must be established.

To our knowledge, six studies examining the validity of self-reported 23vPPV and/or influenza vaccination status have been previously published [5], [6], [7], [8], [9], [10], in addition to one appearing as a conference proceeding [11]. These were based on relatively small numbers of subjects (mean 379, range 135–820), have some data limitations, conflicting results and only two [8], [10] were conducted outside North America. Estimates from earlier studies could have been affected in either direction by poor response rates (as low as 59%) [7], response bias [6] or selection bias [11]. The latest three studies were small (n = 135–354) [8], [9], [10]. None have adjusted estimates for potential confounders except for Zimmerman et al. (restriction of analyses to examine effects of age and vaccination within physicians’ offices) [7] and Mac Donald et al. (time since vaccination with 23vPPV) [6]. Further exploration of the validity of self-reported 23vPPV and influenza vaccination status as an evaluation tool is warranted. Such information can inform the assessment process for the current Australian program in persons aged ≥65 years, as well as similar programs in other countries, and clinicians vaccinating individuals.

As part of a large case–cohort study examining vaccine effectiveness, we undertook to quantify the validity of self-reported vaccination status for 23vPPV and influenza vaccine as an indicator of true vaccination status.

Section snippets

Study design

The case–cohort design is a variant of the case–control study where the comparison group is selected to represent the entire population (the cohort), and is a random sample (independent of case characteristics) of all persons in the population from which the cases are drawn, rather than only those who are unaffected (controls). Hence some cases may also be sampled in the cohort [12]. Unlike a case–control study, no rare disease assumption is required, and the risk ratio (RR) is directly

Description of eligible subjects and data availability

Of 4887 selected subjects, more consented to contacting their vaccine providers to confirm vaccination status (4166/4887, 85%) than to undertaking a questionnaire including self-reported vaccination status (3534/4887, 72%) (Fig. 1). About 2404/3534 (68%) consenting subjects provided a “yes” or “no” response for their 23vPPV status for the preceding 5 years and 3250/3534 (92%) for influenza vaccination status for the previous year. Providers responded to requests for information on vaccination

Influenza vaccine

Estimates from this study of the validity of self-reported influenza vaccination status are consistent with the range provided by previous smaller studies (Table 4) [5], [6], [7], [8], [10]. They remained consistent when the study population was no longer restricted to those providing self-report data in person, first hospital presentations, or where those with “don’t know” responses were assumed to be unvaccinated. Although self-report was highly sensitive, with almost all truly vaccinated

Acknowledgements

The study was approved by the Human Research Ethics Committee, Melbourne Health (ref 2000.022). It was jointly funded by the Victorian Government Department of Human Services and the National Health and Medical Research Council (grant number 146500). The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or preparation or approval of the report. This study was supported by research assistants responsible for data collection: Anne-Marie Woods,

References (18)

There are more references available in the full text version of this article.

Cited by (78)

  • Minimising missed opportunities to promote and deliver immunization services to middle and older age adults: Can hospital-based programs be a solution?

    2021, Vaccine
    Citation Excerpt :

    In 2007, Skull et al found that MOV in hospital was widespread in the 65 years and over cohort [14]. Across two large hospitals over two years, not one of over 2400 patients who required either an influenza or pneumococcal vaccine received it during their selected hospital admission [14]. These trends are comparable to the international literature, which indicates that for adults: a) MOV in hospitals are common, and b) hospital-based vaccination programs are successful in increasing coverage [15–23].

  • Influenza vaccination in the time of COVID-19: A national U.S. survey of adults

    2021, Vaccine
    Citation Excerpt :

    Similarly, like other self-report measures of health behavior, self-report of past influenza vaccination status is subject to recall and social desirability biases. However, there are robust findings that self-report of influenza vaccination status is quite accurate for adults [22–24]. Lastly, intention for future influenza vaccination may not necessarily translate into actual vaccination and may be sensitive to such changes as fluctuations in the status of a viral pandemic, as well as other factors (e.g., the severity of the influenza season).

  • Factors associated with pneumococcal vaccination in 2040 people with type 2 diabetes: A cross-sectional study

    2020, Diabetes and Metabolism
    Citation Excerpt :

    This self-reported information could lead to misclassification, especially if patients had forgotten they had received the vaccine several years ago. Previous studies comparing self-reported pneumococcal vaccination status to medical records have found that the positive predictive value is reasonably high (86%) [20]. Therefore, we considered the risk of misclassification to be low.

View all citing articles on Scopus
View full text