Search strategy and selection criteria
We performed a systematic review and meta-analysis of observational studies assessing unmet healthcare needs. This study followed Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for reporting the manuscript.43 The inclusion criteria were an original article, use of household/community/facility level survey data, use of quantitative analysis, and reporting on outcomes on either forgone care or unmet needs related to healthcare or LTC. Countries at all income levels and all World Health Organization (WHO) regions were included in this study. We excluded qualitative studies, letters, case series, reviews, commentaries, and editorials. Studies based on specific diseases or patient groups were also excluded.
An information specialist did a comprehensive literature search on June 24, 2020, for relevant articles published from inception to June 24, 2020, in the following databases: PubMed, Embase, Web of Science, and CINAHL. No date or language restrictions were applied during the database search. To identify the relevant papers, we combined with “AND” operators of three major topics: (forgone health care OR unmet needs) AND (barrier for healthcare OR long-term care) AND quantitative survey. Further searches for eligible studies were conducted by reviewing references within identified papers. The details of the search strategy are presented in the appendix (Table S1-S4).
Following the study inclusion and exclusion criteria, two independent reviewers first screened the title and abstract (AS, RMM, AM, and RM), and then selected full texts. Any discrepancy among the reviewers during the two stages were resolved through discussions with RM, FG, and ET.
Statistical analysis
A pilot-tested data extraction form was used to collect information from the included articles. Extracted data included the first author’s last name, study country, publication year, survey year, study design, sample size, age range, outcome variables, recall period of outcome variable, barrier framework, explanatory framework, and reasons for unmet needs. We recorded prevalence and event of unmet needs at the overall level and by strata, such as by age, gender, education, occupation, marital status, economic group, migrant status, type of health facilities used, insurance status, geographic location, and type of diseases. Furthermore, we compiled reasons for unmet needs when available. The data extraction form and detailed information of the extracted variables are presented in the appendix (eMethod1).
The primary outcome of interest was unmet healthcare needs and the secondary outcome was unmet needs in LTC. We generally followed the definitions for unmet needs for healthcare or LTC used in the original studies .18,23,25,44−46 In addition, in the present study, we included foregone care, not receiving necessary care, delaying needed medical, dental, or pharmacy care in our definition of unmet healthcare needs. These terms were extracted from key papers and used to corroborate the search strategy used in identifying the original papers. Most studies referring to forgone healthcare measured it by asking questions such as, “Was there a time in the past year you needed a type of (health) care but did not get it?20,47−52 Likewise, studies referring to unmet needs for healthcare measured this by asking questions like, “During the past 12 months, was there ever a time when you felt that you needed health care but didn’t receive it?” 31,34,39,46,53 Although the key word (unmet needs and forgone care) is different, the question actually collects the same information. Therefore, forgone care was included in our definition of unmet healthcare needs. For similar reasons, non-receipt of needed care and delayed care were also included as unmet healthcare needs in our study. The reasons for unmet healthcare needs were derived from survey questions such as, “Thinking of the most recent time (that you didn’t get care when you needed it), why didn’t you get care?” 34,39,53
With regard to defining unmet needs for LTC, previous studies tend to define unmet needs for LTC among older people based on when a person has needs for assistance with ADL or IADL, but the assistance is unavailable, insufficient, or had to wait.15,54 The simplest way to define unmet need is to define the population with LTC needs and assess whether they received assistance. We followed the definition of unmet needs for LTC used in the original studies. Further information about how unmet needs in healthcare and LTC were operationalized in this study are presented in the appendix (eMethod2).
We used prevalence estimates (i.e. proportion of the population with the outcome of interest) for the meta-analysis. When necessary, we calculated prevalence using the original study data provided in the publications. Fixed-effect or random effects meta-analysis was performed depending on the degree of heterogeneity. We used \({I}^{2}\) statistic to assess the level of statistical heterogeneity between the included studies. Based on previous studies, \({I}^{2}\) of < 50 indicated low heterogeneity, between 50% and 75% indicated moderate heterogeneity, and greater than 75% indicated high heterogeneity. We summarized the study-specific estimates using a random-effects model to obtain a pooled prevalence of unmet needs.55 Furthermore, we summarized unmet needs for the following subgroups: the older population (age 65 and above) stratified by reason for unmet needs/barrier dimension, gender (male or female), level of education (primary or less, secondary or college, or higher), self-reported health status (poor/fair, good/average, or very good/excellent), type of illness (NCDs/chronic condition or depression symptoms), insurance enrollment status (insured or uninsured), level of income or socioeconomic status (by quintile, i.e., poorest, poorer, average, rich, or richest), place of residence (urban or rural), and survey year (≤ 2000, 2001–2010, or 2011–2019). All analyses were performed using Stata version 16.1/MP (StataCorp, College Station, TX, USA).