A review of methods to estimate the visibility factor for bias correction in network scale-up studies

Population Size Estimation of People Who Use Illicit Drugs and Alcohol in Iran (2015-2016)

BACKGROUND

Estimating the number of people using illicit drugs and alcohol is necessary for informing health policy and programming. However, it is often challenging to reliably estimate the size of these marginalized populations through direct methods. In this study, we estimated the population size of these groups using the indirect Network Scale-Up (NSU) method in Iran from 2015 to 2016.

METHODS

Using a self-administered questionnaire, we asked 15 124 individuals (54% men) about the number of people they know who used different types of drugs at least once in the past 12 months. Prevalence estimates were reported per 100 000 population. The uncertainty level (UL) was calculated using the bootstrap method.

RESULTS

The average age of the respondents was 33 years old, and 35.1% of them were unmarried. The most common drugs and their prevalence were as follows: opium (2534 [95% UL: 2467-2598]), hashish (849 [95% UL: 811-886]), stimulants (methamphetamine, ecstasy pills, cocaine, and Ritalin) (842 [95% UL: 802-879]), heroin/crack (578 [95% UL: 550-607]), and drug injection (459 [95% UL: 438-484]). Additionally, we estimated the prevalence of alcohol use as 2797 (95% UL: 2731-2861). On average, substance use was 5.23 times more prevalent among men than women. Opium use was more prevalent among individuals aged >50 years old. Moreover, alcohol use was more prevalent among participants between 18 and 30 years old (5164 per 100 000 population).

CONCLUSION

Although opium continues to be the most prevalent illicit drug in Iran, the patterns of illicit drug use are heterogeneous among different age groups, genders, and provinces. Age-gender specific and culturally appropriate interventions are warranted to meet the needs of people in different subgroups.

Estimating the Size of Hidden Groups

NSU studies have been applied in different countries. For example, this technique has been used in the United States to estimate the number of women who had been raped in the last year.

Thirty Years of The Network Scale-up Method

Estimating the size of hard-to-reach populations is an important problem for many fields. The Network Scale-up Method (NSUM) is a relatively new approach to estimate the size of these hard-to-reach populations by asking respondents the question, "How many X's do you know," where X is the population of interest (e.g. "How many female sex workers do you know?"). The answers to these questions form Aggregated Relational Data (ARD). The NSUM has been used to estimate the size of a variety of subpopulations, including female sex workers, drug users, and even children who have been hospitalized for choking. Within the Network Scale-up methodology, there are a multitude of estimators for the size of the hidden population, including direct estimators, maximum likelihood estimators, and Bayesian estimators. In this article, we first provide an in-depth analysis of ARD properties and the techniques to collect the data. Then, we comprehensively review different estimation methods in terms of the assumptions behind each model, the relationships between the estimators, and the practical considerations of implementing the methods. We apply many of the models discussed in the review to one canonical data set and compare their performance and unique features, presented in the supplementary materials. Finally, we provide a summary of the dominant methods and an extensive list of the applications, and discuss the open problems and potential research directions in this area.

Using Social Networks to Estimate the Number of COVID-19 Cases: The Incident (Hidden COVID-19 Cases Network Estimation) Study Protocol

Recent literature has reported a high percentage of asymptomatic or paucisymptomatic cases in subjects with COVID-19 infection. This proportion can be difficult to quantify; therefore, it constitutes a hidden population. This study aims to develop a proof-of-concept method for estimating the number of undocumented infections of COVID-19. This is the protocol for the INCIDENT (Hidden COVID-19 Cases Network Estimation) study, an online, cross-sectional survey with snowball sampling based on the network scale-up method (NSUM). The original personal network size estimation method was based on a fixed-effects maximum likelihood estimator. We propose an extension of previous Bayesian estimation methods to estimate the unknown network size using the Markov chain Monte Carlo algorithm. On 6 May 2020, 1963 questionnaires were collected, 1703 were completed except for the random questions, and 1652 were completed in all three sections. The algorithm was initialized at the first iteration and applied to the whole dataset. Knowing the number of asymptomatic COVID-19 cases is extremely important for reducing the spread of the virus. Our approach reduces the number of questions posed. This allows us to speed up the completion of the questionnaire with a subsequent reduction in the nonresponse rate.

Novel approaches to estimating abortion incidence

Background

In countries where abortion is legally restricted or clandestine, estimates of abortion incidence are needed in order to bring attention to the reality of this practice. Innovations in methods for estimating stigmatized behaviors, coupled with changes in the conditions under which women obtain abortions, prompt us to review new approaches to estimating abortion incidence and propose innovations in this field.

Methods

We discuss five approaches for yielding accurate estimates in countries with restrictive abortion laws. These include two prevailing approaches in the field (direct questioning of women about their abortions and the Abortion Incidence Complications Method (AICM)), one that has begun to be in use in recent years (the List Experiment) and two that are newly proposed by the authors (the Confidante Approach and a modification of the AICM). We discuss assumptions, strengths and weaknesses of each approach. Finally, we suggest strategies for assessing the validity of the findings in the absence of a gold standard.

Results

Though direct questioning has consistently been shown to miss many abortions, reporting can potentially be improved by normalizing or reframing the experience of abortion. The AICM has had the advantage of not relying on women’s reports about their abortions; however as self-induced abortion becomes more common, this strength becomes a weakness. The modified AICM, which uses women’s abortion reports to estimate the proportion of abortions that lead to treated complications, improves our chances of capturing self-induced abortions. The List Experiment preserves the woman’s anonymity (not just her confidentiality), but it can be cognitively challenging and the potential to make subgroup estimates is extremely limited. The Confidante Approach entails asking survey respondents about abortions among women who confide in them, rather than their own abortions. An adjustment factor can be applied to estimate the incidence of confidantes’ abortions that are unknown to respondents. This approach relies on the assumption that women know and will report whether their confidantes had an abortion.

In the absence of a gold standard measure of abortion incidence, four strategies can be employed to compare and assess these approaches: (a) comparing the level of underreporting across methods susceptible to underreporting but not to overreporting, (2) validating components of abortion estimates against an objective measure, (3) testing whether these strategies accurately estimate other sensitive behaviors for which a gold standard exists, and 4) sensitivity analyses. Ultimately, it might be appropriate to employ more than one methodology when measuring abortion incidence.