Estimating the Size of Most-at-Risk Populations for HIV/AIDS Using the Generalized Network Scale-Up Method: A Study in Shiraz,IranIran

Abstract

Accurate estimation of the sizes of most-at-risk populations (MARPs) for HIV infection is critical for effective intervention planning, resource allocation, and service coverage assessment. This study aimed to estimate the population sizes of female sex workers (FSWs), people who inject drugs (PWIDs), and men who have sex with men (MSM) in the 18–45-year-old population of Shiraz, Iran, using the network scale-up (NSU) method. A cross-sectional study was conducted involving 768 purposively selected individuals aged 18–45 years residing in Shiraz for at least 5 years. Data were collected through a self-administered questionnaire, utilizing 16 reference groups with known frequencies to estimate average social network size. The prevalence of MARPs was then estimated and adjusted using established correction factors. The average active social network size was 203 contacts (103 males and 100 females). Adjusted size estimations indicated the highest prevalence among FSW at 2,181 per 100,000 (95% confidence interval [CI]: 1,979–2,375), followed by PWID at 561 per 100,000 (95% CI: 488–633) and MSM at 418 per 100,000 (95% CI: 272–563). Age distribution analyses revealed the greatest PWID prevalence in the 35–39 age group, FSW in 30–34, and MSM in 25–29. The estimated sizes of these populations substantially exceeded current harm-reduction program coverage in the region. Application of the NSU method revealed significant gaps between the actual sizes of high-risk populations and those reached by existing interventions in Shiraz. Targeted expansion of prevention, testing, and treatment programs, alongside initiatives to combat stigma and improve outreach, is urgently needed to control HIV transmission among these hidden and vulnerable groups.

Keywords

size estimation indirect methods network scale-up most-at-risk populations HIV

Introduction

In the Middle East and North Africa region, HIV infection is predominantly concentrated among subgroups exhibiting high-risk behaviors, including people who inject drugs (PWIDs), female sex workers (FSWs) and their clients, and men who have sex with men (MSM).^1,2 These groups are classified as the most-at-risk populations (MARPs) for HIV. In countries such as Iran, where the HIV epidemic is concentrated, these subgroups play a pivotal role in the transmission of the virus to the general population.^2–5

To effectively plan interventions, allocate resources, estimate the number of infected individuals, predict disease burden, measure service coverage, and evaluate health care services, reliable population estimates of these groups are essential at both local and national levels.^6–9 Accurate measurement of the sizes of these populations is a critical component of HIV infection control, as highlighted in the Global CDC/UNAIDS and WHO HIV/AIDS Surveillance Meeting in Bangkok in March 2009.⁹ However, due to the nature of their behaviors, along with stigmatization, discrimination, and fear of legal punishment, these groups remain hidden and are considered hard-to-reach populations. Consequently, accurate estimation of their sizes poses a significant challenge within national HIV/AIDS surveillance systems.^2,5,8,10,11

Standard sampling methods and direct estimation techniques are not recommended for estimating the sizes of these hidden populations.^12,13 Recent years have seen the development of various techniques and guidelines for measuring the sizes of these groups, including enumeration, census, capture-recapture, and network scale-up (NSU) methods. Each of these methods has its own advantages and disadvantages.^6,9,14 Among them, NSU is one of the most significant and widely utilized methods, as it does not require direct contact with high-risk groups and allows for the estimation of multiple groups in a single study.^11,14 However, this method is also subject to certain errors.^15,16 The present study aims to estimate the sizes of MARPs for HIV among the 18–45-year-old population of Shiraz using the NSU method, while also refining existing errors.

Materials and Methods

Sampling and data collection

This cross-sectional study aimed to estimate the sizes of MARPs for HIV infection in Shiraz, the capital of Fars Province, utilizing the NSU method. The target population consisted of individuals aged 18–45 years who had resided in Shiraz for a minimum of 5 years. According to the latest National Population and Housing Census conducted by the Statistical Center of Iran, the total population of Shiraz is approximately 1,712,745, with 866,160 individuals falling within the 18–45 age range. A total of 768 eligible participants were interviewed using a self-administered questionnaire.

The requirement of at least 5 years of residence in Shiraz was applied to ensure comparability with previous NSU studies conducted in Iran and to increase the likelihood that respondents had developed stable and representative social networks within the study setting.^8,10,14,17 Individuals with shorter residency may have smaller or less stable and geographically dispersed networks, which could introduce additional bias and increase transmission error in NSU-based estimations.

Participants were selected through purposive sampling from the 12 regional districts of Shiraz. Only those who provided written informed consent were included in the study. Given the limitations of random sampling in obtaining a representative sample of the target population, quota sampling was employed based on sex (50% male and 50% female) and socioeconomic status, with four regional districts allocated to each socioeconomic category.

To minimize sensitivity regarding certain questions, interviews were conducted by gender-matched trained interviewers in public areas such as parks and streets. Participants were informed about the study’s objectives and the voluntary nature of their participation, after which they signed informed consent forms approved by Shiraz University of Medical Sciences. If an individual declined to participate, a replacement with similar characteristics was recruited. The study questionnaire was divided into three sections: (1) demographic information, (2) questions designed to estimate the average social network size within the target population, and (3) questions aimed at estimating the frequency of individuals belonging to MARPs within the 18–45 age group.

NSU method

The NSU method is predicated on the assumption that the prevalence of target groups within a representative social network mirrors their prevalence in the overall population. In this method, respondents are queried about the high-risk behaviors of individuals within their social networks, which mitigates the risks of underreporting and underestimating sensitive and stigmatized behaviors. The primary formula utilized in the NSU method is as follows:

\frac{m}{c} = \frac{e}{t}

where

(m) represents the number of individuals known by the respondent in the target group (e.g., the number of PWID in the respondent’s network),

(e) is the estimated size of the target group, and

(t) is the total size of the target population.

Accordingly, the average size of the individuals’ social network can be computed using the following formula:

c_{i} = t * \frac{\sum_{j = 1}^{L} m_{i j}}{\sum_{j = 1}^{L} e_{j}}

where j and i represent the respondent’s group and the respondent, respectively.

In this study, the term “active social network” refers to the number of individuals with whom the respondent has direct contact, whether in person, via telephone, email, or any internet-based communication platforms (including social media such as Instagram, Snapchat, Discord, and similar services), and can connect with if necessary.

Calculation and refinement of C

To estimate the average social network size, we selected 16 first names (eight masculine and eight feminine) with known prevalence rates ranging from 0.1% to 4%. Names that were gender-neutral or combined were excluded. The frequency of the selected names was obtained from the National Organization for Civil Registration. We aggregated the data from all 16 names to calculate the value of (C), which was subsequently used to back-calculate the sizes of reference groups. The ratio of the actual size to the back-calculated size was computed for each name. Any ratio outside the range of 0.5–2 was deemed ineligible. During each iteration of the (C) calculation, the reference group with the least favorable ratio was excluded, and (C) was recalculated using the remaining reference groups. This process continued until all ratios fell within the acceptable range. Ultimately, the sizes of the MARPs were determined based on the final (C) value.

Crude estimation of size of MARPs

Using the final (C) value, the size of the target group was computed through the NSU method as follows:

e = t * \frac{\sum_{j = 1}^{L} m_{i j}}{\sum_{i = 1}^{L} c_{i}}

We defined MARPs as individuals who engaged in the specified high-risk behaviors at least once in the past year. For example, an individual who injected drugs even once during that period was classified as a member of the PWID group.

Generalized NSU

Two primary sources of error affect the accuracy of the NSU method: transmission error and relative network size bias. The NSU method assumes that respondents possess complete awareness of the behaviors of individuals in their social networks. Transmission error occurs when respondents are unaware of the sensitive or stigmatized behaviors of individuals within their social network, leading to underreporting of the target population. To address this, a visibility factor is applied to correct for the probability that the behavior is not visible or known to the respondent.

The NSU method assumes that respondents possess complete awareness of the behaviors of individuals in their social networks. Relative network size bias arises because members of high-risk populations often have smaller social networks compared with the general population, which can distort size estimates. This bias is corrected using a popularity factor, which adjusts for differences in social network sizes between target groups and the general population.

By incorporating these correction factors, the generalized NSU method enhances the accuracy of population size estimations for hidden and hard-to-reach groups, ensuring more reliable data for public health planning and intervention.

We used the correction factors from the study conducted by Maghsoudi et al., in which adjustment coefficients for the two main sources of error in the NSU method—transmission error and relative network size bias—were calculated specifically for estimating the size of MARPs for HIV infection.¹⁶

Finally, the size of the target group was computed using the following formula:

\hat{e} = t * \frac{\sum_{j = 1}^{L} m_{i j}}{\sum_{i = 1}^{L} c_{i}} \times \frac{1}{P F} \times \frac{1}{V F}

The 95% uncertainty intervals (UIs) of C and e were estimated by applying the bootstrap procedure using 1,000 iterations.

Results

From 768 participants, the sample was evenly split by sex (50% male and 50% female). The mean age was 30.11 years with a standard deviation of 8.57 years. Academic education was reported by 60% of males and 55% of females. Additional demographic details are provided in Table 1.

Table 1.

The Participants’ Demographic Characteristics

Variable	Category	Frequency (percentage)
Age (years)	18–24	247 (32.2)
	25–29	142 (18.5)
	30–34	127 (16.5)
	35–39	95 (12.4)
	40–45	157 (20.4)
Education level	Ability to read and write	58 (7.6)
	High school diploma	268 (34.9)
	Academic	442 (57.6)
Socioeconomic status	High	256 (33.3)
	Moderate	256 (33.3)
	Low	256 (33.3)

After computing the ratio of actual to back-calculated values, two reference groups were excluded from the final C estimates. The average active social network for individuals aged 18–45 years in Shiraz was found to be 203 contacts, comprising 103 males and 100 females. Accordingly, each individual aged 18–45 years in Shiraz knew an average of 203 people within the same age range (103 males and 100 females). The network sizes across different demographic strata are presented in Table 2.

Table 2.

Social Network Size Based on Participants’ Demographic Variables

Variable	Subgroups	Network size	SD of network size
Sex	Male	200	44.3
Sex	Female	205	44.02
Age group	18–24	200	38.9
	25–29	195	42.9
	30–34	203	44.5
	35–39	197	44.9
	40–45	216	49.4
Education level	Ability to read and write	222	57.9
	High school diploma	205	45.1
	Academic	199	40.9
Socioeconomic status	High	199	41.6
	Moderate	205	45.8
	Low	204	45.0

SD, standard deviation.

Across the study, the total number of PWIDs, FSWs, and MSM known by participants was 976. The active social network consisted of 326 PWID, 584 FSW, and 66 MSM.

According to the most recent National Population and Housing Census of Iran, 866,160 individuals aged 18–45 years reside in Shiraz, including 429,555 males and 436,605 females. The crude and adjusted MARP sizes are shown in Tables 3 and 4, respectively. Tables 3 and 4 indicate that the highest MARP frequency was among FSW, while the lowest was among MSM.

Table 3.

Crude Size Estimation of MARPs

MARPs	Σmi	Σci	t	e
PWID	326	155,805.75	866,160	1,812.30
FSW	584	77,118.94	436,605	3,306.29
MSM	66	79,036.18	429,555	358.70

FSW, female sex worker; MARP, most-at-risk population; MSM, men who have sex with men; PWID, people who inject drug.

Table 4.

Adjusted Size Estimation of MARPs

MARPs	Crude estimation	Popularity factor	Visibility factor	Adjusted estimation (95% CI)	Frequency in 100,000(95% UI)
PWID	1,812.30	0.70	0.54	4,857 (4,231–5,483)	561 (488–633)
FSW	3,306.29	0.76	0.44	9,522 (8,642–10,370)	2,181 (1,979–2,375)
MSM	358.70	1	0.2	1,794 (1,168.50–2,419)	418 (272–563)

CI, confidence interval; FSW, female sex worker; MARP, most-at-risk population; MSM, men who have sex with men; PWID, people who inject drug; UI, uncertainty interval.

Crude and adjusted MARP sizes by age groups are shown in Table 5.

Table 5.

Crude and Adjusted Size Estimation of MARPs Based on Age Groups

Age group		Σmi	Σci	t	Crude e	Adjusted e	Rate in 100,000
18–24	PWID	23	49,464.10	176,534	82.08	219.97	125
	FSW	94	24,221.92	86,817	336.91	970.30	1,118
	MSM	29	25,417.46	89,717	93.76	468.80	523
25–29	PWID	32	27,746.65	182,977	211.02	565.53	309
	FSW	132	13,969.92	93,803	886.33	2,552.63	2,721
	MSM	19	13,780.76	89,174	122.94	614.70	689
30–34	PWID	56	25,870.21	206,486	446.97	1,197.87	580
	FSW	192	13,290.15	104,926	1,515.84	4,365.61	4,161
	MSM	9	12,518.53	101,560	73.01	365.05	359
35–39	PWID	125	18,672.08	159,478	1,067.62	2,861.22	1,794
	FSW	107	8,961.83	80,629	962.67	2,772.48	3,439
	MSM	6	9,821.17	78,849	48.17	240.85	305
40–45	PWID	90	34,052.71	140,685	371.82	966.47	687
	FSW	59	16,675.11	70,430	249.19	717.66	1,019
	MSM	3	17,498.28	70,255	12.04	60.20	86

FSW, female sex worker; MARP, most-at-risk population; MSM, men who have sex with men; PWID, people who inject drug.

The results indicate that the highest frequencies of PWID, FSW, and MSM occur in the 35–39, 30–34, and 25–29 age groups, respectively. Conversely, the lowest frequency of PWID is in the 18–24 age group, while the lowest frequencies for FSW and MSM are in the 40–45 age group.

Discussion

Active social network

Based on comparisons across different methodologies, the present study employed the known population approach to estimate the size of active social networks. Sixteen first names were used as reference groups. Previous research has confirmed the high accuracy of this method for estimating social network size.¹⁰

In our study, the average active social network size among individuals aged 18–45 years in Shiraz was 203 persons (103 males and 100 females). Reported estimates vary widely due to differences in target populations and methodological approaches. For example, Killworth reported a network size of 286 using the known population method.¹⁸ In China, Gue et al. applied backward estimation across 19 reference groups and found an average of 310 for adults aged 18–60 years.¹⁹ In Iran, the mean active social network size was estimated at 308 among individuals over 18 years of age.²⁰ Such discrepancies may arise from differences in sampling strategies, geographic coverage, and age groups under study. Whereas the cited studies were conducted nationally, our investigation focused on a single city with a narrower age range, which likely explains the smaller estimate. Shati et al. reported a C value of 259 in Tehran for individuals >18 years,²¹ which is close to our estimate. This similarity may reflect the restriction of participants’ networks to a single city, while minor differences could be explained by age range, geographic, cultural, and demographic variations.

Size estimation of MARPs

People who inject drugs

Several studies have estimated the size of the population of PWID in Iran. In this study, the prevalence of PWID among 18–45-year-olds in Shiraz was 561 per 100,000 (0.5%). Shokoohi et al. reported a prevalence of 1.2% among the same age group in Kerman,¹⁴ more than twice our estimate. This difference may be partly explained by gender composition, as most PWIDs in Iran are male, while our estimate included both sexes. Jafari et al. found a prevalence of 280 per 100,000 using the NSU method,¹⁰ approximately half of our estimate. The higher estimate in our study may be related to age distribution, as injection drug use is less common in those <18 and >45 years.²²

Narouee et al. estimated a prevalence of 1,263 per 100,000 among adults in Iranshahr, Sistan, and Baluchistan province,⁸ about twice the prevalence we observed. This region has socioeconomic vulnerabilities such as high unemployment and proximity to drug-trafficking routes, which contribute to higher rates of injection drug use. In Jahrom, Fars province, the NSU method yielded a prevalence of 0.26% among males aged 18–45 years.¹⁷ The lower estimate may reflect the absence of correction for NSU-related biases, potentially leading to underestimation.

Female sex workers

Few studies have assessed the size of FSW in Iran. In the present study, the prevalence of FSW among women aged 18–45 years in Shiraz was 2,180 per 100,000 (about 2%). This indicates that nearly 2% of women in this age group reported at least one transactional sexual relationship in the past year. Sharifi et al. estimated the prevalence of FSW across 13 Iranian cities using the NSU method in 2014. Prevalence ranged from 0.14% in Tabriz to 2.44% in Isfahan, with 10 cities reporting rates between 1% and 2% and a national average of 1.54%.⁵ In Shiraz, their study estimated a prevalence of 1.67% among women aged 15–49 years,⁵ which is broadly consistent with our findings. Slight differences may reflect variations in age groups.

By contrast, Jafari et al. reported a prevalence of 709 per 100,000 among women over 18 years in Tabriz,¹⁰ roughly one-third of our estimate. Similarly, Sharifi’s study showed the lowest national prevalence in Tabriz.⁵ Such discrepancies likely reflect differences in age ranges, socioeconomic status, and sociocultural conditions between cities.

Men who have sex with men

Due to cultural sensitivities, stigma, and discrimination, few studies have assessed the size of MSM in Iran. Shokoohi estimated a prevalence of 0.5% among men aged 18–45 years in Kerman (95% UI: 0.2%–1%),¹⁴ consistent with our estimate of 0.42%. This similarity may reflect comparable target populations and methodologies.

Jafari reported a prevalence of 247 per 100,000 among men over 18 years in Tabriz,¹⁰ lower than our estimate of 420 per 100,000. Age distribution may account for this difference, as MSM are predominantly younger than 30. Regional sociocultural variations may also contribute. Hemayatkhah reported a prevalence of 0.34% (95% UI: 0.17%–0.85%) among men aged 18–45 years in Jahrom,¹⁷ which is in line with our findings, again reflecting methodological and demographic similarities.

Implications for policy and practice

Our findings reveal that the estimated sizes of PWID, FSW, and MSM in Shiraz substantially exceed the current coverage of existing harm-reduction programs. In Shiraz, these services are primarily delivered through fixed and outreach facilities, including drop-in centers offering needle and syringe programs (NSPs), methadone maintenance therapy, and voluntary counseling and testing for HIV.^23,24 While these programs exist, their coverage remains insufficient. Official reports suggest that current services reach only a fraction of the estimated MARPs.^25,26 Previous research among PWID in Iran has documented that while awareness of NSPs reached 62.8%, actual utilization was reported by only 54.8% of this population. This awareness-to-uptake gap is even more pronounced for drug treatment services, with awareness at just 19.7% and utilization at a mere 9.1%.²⁶ Utilization is further hampered by profound stigma, fear of legal repercussions, and a lack of awareness about available services.^1,23,24 This substantial gap between the estimated population size and the reach of current programs underscores the urgent need to expand services, implement mobile units, and integrate HIV services with other health care provisions to reduce stigma and improve access for these marginalized groups.

Limitations

This study has a limitation that should be considered when interpreting the findings. The potential transiency of MARPs may have influenced the accuracy of our population size estimates. Due to the stigmatized and often illegal nature of their behaviors, individuals in these groups frequently experience residential instability or migrate to larger cities to avoid legal consequences, seek employment opportunities, or access specialized services.^27–30 Our estimates were based on a 5-year residency criterion, which may have captured a more stable subset of these populations while underestimating the total number of individuals who temporarily reside in or pass through Shiraz. Future studies should incorporate data on migration patterns and residential history to better understand the population dynamics of MARPs and to inform the design of targeted, flexible, and mobile health interventions.

Conclusion

This study applied the generalized NSU method to estimate the population sizes of the most-at-risk groups for HIV infection—namely FSW, PWID, and MSM—among adults aged 18–45 years in Shiraz, Iran. The findings revealed that the estimated sizes of these high-risk groups are substantially greater than the current coverage of harm-reduction programs. These results highlight the persistent gap in service coverage and underscore the urgent need to expand targeted HIV prevention and treatment initiatives. Implementing comprehensive strategies for increasing access to health care, reducing stigma, and strengthening surveillance measures is critical for controlling the HIV epidemic among hidden and marginalized populations. Policymakers and health authorities should utilize these robust estimates to inform resource allocation, develop more inclusive harm-reduction programs, and monitor progress toward national and regional HIV/AIDS control goals.

Ethical Approval

The study protocol was approved by the ethics committee of Shiraz University of Medical Sciences [Ethical code: (IR.SUMS.REC.1398.217)]. Before obtaining verbal informed consent, participants were informed about the anonymity of all collected data and the voluntary nature of their participation. Data were reported as aggregate data and did not identify subjects as individuals.

Data Availability

Data supporting the findings of this study are available from the corresponding author upon reasonable request. Access to restricted data may be provided under data-sharing agreements approved by the authors’ institution and relevant ethics boards.

Authors’ Contributions

All authors contributed significantly to the development of this research. J.H. conceptualized and designed the study and led the article writing. A.M. and M.R.B. assisted with data collection and data synthesis and contributed to writing and critical revisions of the article. All authors reviewed and approved the final version of the article and agreed to be accountable for all aspects of the work.

Footnotes

Acknowledgments

The authors would like to thank Ms. A. Keivanshekouh at the Research Improvement Center of Shiraz University of Medical Sciences for improving the use of English in the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported by the Research Vice-chancellor of Shiraz University of Medical Sciences under grant (97-01-04-17410).

References

Fahimfar

, Sedaghat

, Hatami

, Kamali

, Gooya

. Counseling and harm reduction centers for vulnerable women to HIV/AIDS in Iran. Iran J Public Health, 2013; 42(Supple1):98–104.

Karami

, Khazaei

, Poorolajal

, Soltanian

, Sajadipoor

. Estimating the population size of female sex worker population in Tehran, Iran: Application of direct capture–recapture method. AIDS Behav, 2017; 21(8):2394–2400.

Khodadost

, Ghalehkhani

, Hajebi

, et al. Hot spot mapping and size estimation of illicit drug users in Iran: A pilot study. 2021; doi: 10.21203/rs.3.rs-141437/v1

Rahimi-Movaghar

, Amin-Esmaeili

, Haghdoost

A-A

, Sadeghirad

, Mohraz

. HIV prevalence amongst injecting drug users in Iran: A systematic review of studies conducted during the decade 1998–2007. Int J Drug Policy, 2012; 23(4):271–278.

Sharifi

, Karamouzian

, Baneshi

, et al. Population size estimation of female sex workers in Iran: Synthesis of methods and results. PLoS One, 2017; 12(8):e0182755.

Arumugam

, Kangusamy

, Sahu

, Adhikary

, Kumar

, Aridoss

. Size estimation of high-risk groups for HIV infection in India based on data from national integrated bio-behavioral surveillance and targeted interventions. Indian J Public Health, 2020; 64(Supplement):S39–S45.

Jing

, Qu

, Yu

, Wang

, Cui

. Estimating the sizes of populations at high risk for HIV: A comparison study. PLoS One, 2014; 9(4):e95601.

Narouee

, Shati

, Nasehi

, Dadgar

. The size estimation of injection drug users (IDUs) using the network scale-up method (NSUM) in Iranshahr, Iran. Med J Islam Republ Iran, 2019; 33:158.

Saidel

, Loo

, Salyuk

, Emmanuel

, Morineau

, Lyerla

. Applying current methods in size estimation for high risk groups in the context of concentrated epidemics: Lessons learned. jHASE-J HIV AIDS Surveill Epidemiol, 2010; 2(1).

10.

JafariKhounigh

, Haghdoost

, SalariLak

, et al. Size estimation of most-at-risk groups of HIV/AIDS using network scale-up in Tabriz, Iran. Journal of Clinical Research & Governance, 2014; 3(1):21–26.

11.

Maghsoudi

, Jalali

, Neydavoodi

, Rastad

, Hatami

, Dehghan

. Estimating the prevalence of high-risk behaviors using network scale-up method in university students of Larestan in 2014. J Subst Use, 2017; 22(2):145–148.

12.

Laga

, Bao

, Niu

. Thirty years of the network scale-up method. J Am Stat Assoc, 2021; 116(535):1548–1559; doi: 10.1080/01621459.2021.1935267

13.

Zamanian

, Baneshi

, Haghdoost

, Mokhtari-Sorkhani

, Amiri

, Zolala

. Estimating the size and age-gender distribution of women’s active social networks. Addict Health, 2016; 8(3):170–178.

14.

Shokoohi

, Baneshi

, Haghdoost

A-A

. Size estimation of groups at high risk of HIV/AIDS using network scale up in Kerman, Iran. Inte J Prev Med, 2012; 3(7):471.

15.

Bernard

, Hallett

, Iovita

, et al. Counting hard-to-count populations: The network scale-up method for public health. Sex Transm Infect, 2010; 86(Suppl_2):ii11–ii15.

16.

Maghsoudi

, Baneshi

, Neydavoodi

, Haghdoost

. Network scale-up correction factors for population size estimation of people who inject drugs and female sex workers in Iran. PLoS One, 2014; 9(11):e110917.

17.

Hemayatkhah

, Rahmanian

, Haghdoost

. Population size estimation of groups at high risk of HIV/AIDS in men, using network scale up in South of Iran, 2017. Journal of Isfahan Medical School, 2019; 36(512):1662–1669.

18.

Killworth

, McCarty

, Bernard

, Shelley

, Johnsen

. Estimation of seroprevalence, rape, and homelessness in the United States using a social network approach. Eval Rev, 1998; 22(2):289–308.

19.

Guo

, Bao

, Lin

, et al. Estimating the size of HIV key affected populations in Chongqing, China, using the network scale-up method. PLoS One, 2013; 8(8):e71796.

20.

Rastegari

, Haji-Maghsoudi

, Haghdoost

, Shatti

, Tarjoman

, Baneshi

. The estimation of active social network size of the Iranian population. Glob J Health Sci, 2013; 5(4):217–227.

21.

Shati

, Haghdoost

, Majdzadeh

, Mohammad

, Mortazavi

. Social network size estimation and determinants in tehran province residents. Iran J Public Health, 2014; 43(8):1079–1090.

22.

Nikfarjam

, Shokoohi

, Shahesmaeili

, et al. National population size estimation of illicit drug users through the network scale-up method in 2013 in Iran. Int J Drug Policy, 2016; 31:147–152.

23.

Ekhtiari

, Noroozi

, Farhoudian

, et al. The evolution of addiction treatment and harm reduction programs in Iran: A chaotic response or a synergistic diversity? Addiction, 2020; 115(7):1395–1403.

24.

Zafarghandi

MBS

, Jadidi

, Khalili

. Iran’s activities on prevention, treatment and harm reduction of drug abuse. Int J High Risk Behav Addict, 2015; 4(4):e22863.

25.

Ghalekhani

, Hajebi

, Ghodusi

, et al. Programmatic mapping and size estimation of people who inject drugs to plan targeted HIV prevention and harm reduction programs in Iran, a nationwide study. BMC Public Health, 2025; 25(1):2428.

26.

Rahnama

, Mohraz

, Mirzazadeh

, et al. Access to harm reduction programs among persons who inject drugs: Findings from a respondent-driven sampling survey in Tehran, Iran. Int J Drug Policy, 2014; 25(4):717–723.

27.

Arum

, Fraser

, Artenie

, et al.; Homelessness, HIV, and HCV Review Collaborative Group. Homelessness, unstable housing, and risk of HIV and hepatitis C virus acquisition among people who inject drugs: A systematic review and meta-analysis. Lancet Public Health, 2021; 6(5):e309–e23.

28.

Grieb

SMD

, Davey-Rothwell

, Latkin

. Housing stability, residential transience, and HIV testing among low-income urban African Americans. AIDS Educ Prev, 2013; 25(5):430–444.

29.

Goldenberg

, Strathdee

, Perez-Rosales

, Sued

. Mobility and HIV in Central America and Mexico: A critical review. J Immigr Minor Health, 2012; 14(1):48–64.

30.

Goldenberg

, Silverman

, Engstrom

, Bojorquez‐Chapela

, Strathdee

. “Right Here is the Gateway”: Mobility, sex work entry and HIV risk along the Mexico–US Border. Int Migr, 2014; 52(4):26–40.