Evaluatiing the Effects of HIV Self-testing used Professional Rapid Test Kits among MSM in Some Areas of China.

With rapid expanding of HIV self-testing (HIVST) among unprofessional people has many unknown hidden dangers. This research for the rst to survey the feasibility of urine, oral mucosal transudate (OMT) and nger blood rapid HIV testing kits were used for HIVST by non-professionals MSM in China. Total 274 valid questionnaires were received from 313 MSM participants, including 263 completed urine HIVST, 61 completed OMT HIVST and 17 completed blood HIVST. The average age of participants was under 30, about 80% were unmarried, more than 80% with an education level above Grade 9 but more than 50% people had never heard of the rapid HIV test. There were signicant differences in the key information understanding accuracy between HIVST. The accuracy rates were 18.0–80.6%. When the HIVST result was positive, more than 80% chose to seek conrmation. When the test was negative, 60.5% (159/263) participants of urine self-test and 32.8% (20/61) of OMT self-test chose regular retesting. When the test was ineffective, more than 80% chose to retest. In addition, 54.1% of the 146 voluntary participants to accept blood HIVST, followed by 15.8% accept blood and urine HIVST and 14.4% accept only urine HIVST. The main reason for choosing blood HIVST was "accuracy", while the main reason for choosing urine HIVST was "convenience". This ndings lays a scientic theoretical basis for further carring out HIVST in China.

speci cations for WHO prequali cation of Human Immunode ciency Virus (HIV) rapid diagnostic tests for professional use and/or self-testing" [7], we evaluated the understanding ability of key information in manual, the interpretation ability of simulation results, and the operation ability of HIVST. At the same time, we also evaluated the preference of the people for urine, blood and OMT. In order to lays a scienti c theoretical basis for further carring out HIVST in China.
Additionally, the comprehension correct rates of urine, OMT and blood rapid test results were 62.4%~79.5%, 57.4%~88.5% and 17.7%~88.2%, respectively. The most accurate understand was "if there is no C line and no test line, it is invalid ", while the inaccurate understand were "if there is no C line, it is invalid" and "if there is no C line and the test line appear, it is invalid". Especially for the interpretation of "HIV infection cannot be excluded when the test is negative" and "HIV infection cannot be con rmed when the test is positive", the comprehension correct rate both less than 50% (Table 2).

Results analog pictures discriminant evaluation
The participants had good interpretation accuracy (86.9%~98.1%) with the results analog pictures of strong reactivity, non-reactivity and invalid (no quality control line, no test line), but had low accuracy (32.7%~68.9%) with the simulation pictures of weak reactivity and invalid (no quality control line, no test line) results.

HIVST operation evaluation
The urine HIVST performance with the highest accuracy during the self-detection evalution. 82.1% (216/263) participants could perform all the urine HIVST steps correctly, 11.0% (29/263) had an operation error of one step and 6.9% (18/263) had multi-step operation error. The "add 3 drops urine to the sample area in the test card" with lowest correct rate was 88.2% (232/263) ( Table 3).
The operating accuracy of OMT and blood HIVST were 11.5% (7/61) and 23.5% (4/17), respectively. Especially with the low operate accury in "slightly gargling with warm water before collection" and "two wiping time 5-6s" of OMT HIVST, "add 1 drop of whole blood to the add sample hole" and "read the result after 30 minutes" of blood HIVST (Table 3).

Discussion
Non-professionals using professional rapid HIV testing reagents for HIVST may have problems. The stages of sample collection, testing and result interpretation[8-10] may affect the test accuracy. In this study, MSM participants were asked to do HIVST using rapid HIV testing kits without any assistance or guidance. The results showed that MSM can better complete urine HIVST with the operation accuracy rate is 82.1%, which is consistent with the research of Roger B. Peck et al [11]. However, the operating errors were common in OMT and blood HIVST, the operating accuracy rate were only 11.5% and 23.5%, respectively. Studies have shown that the high error rate in OMT HIVST is due to the complex OMT collection and test process, the too numerous descriptions in the instructions to users' failure to read or understand carefully [12]. And the most common error in blood HIVST was "reading results after 30 minutes", perhaps due to participants were anxious to know their infection status and immediately interpreted the results when the test strip display. This is different from the study of Mohammed Majam and Smith P [13,14], which operational errors mainly occurred in the blood collection and sampling process. In addition, during the urine HIVST, the most mistakes were "adding 3 drops of urine to the sample area of the test card", which may be due to some participants did not read the instructions carefully. Therefore, this study suggest that the clear, understandable and interesting instructions should be used to improve the feasibility of non-professionals HIVST [11].
Results analog picture discriminant evaluation showed that the strong reactive and non-reactive results with the highest interpretation accuracy, while the weakly reactive results with low interpretation accuracy, which was consistent with the relevant research results [9,11,12,15]. Especially, the more complex invalid results of "no quality control line /T line, with test line /C line" is far lower than the invalid results of "no quality control line /T line, no test line /C line". In this regard, relevant studies have shown that use different symbols for test lines and control lines could increase the results interpretation accuracy [11].
Therefore, it is suggested that clear, simple and easy to distinguish symbols can improve the results interpretation accuracy.
According to the evaluation results of participants' understanding the cautions, operation processes and results interpretation in the reagent instructions, more than half questions with less than 70% comprehension accuracy rate, which was consistent with the relevant domestic studies [12]. However, the study of Gresenguet [15] showed most people could correctly understand the instruction information. It may be the reagent instructions used in this study were full of words and contained technical terms, which led to participants' inability to accurately identify and effectively understand while simple and understandable colloquial descriptions of instructions were used in Gresenguet's study. More important, the low understanding accuracy of "HIV infection cannot be ruled out if the test result is negative, and HIV infection cannot be con rmed if the test result is positive" will affect the users' subsequent solutions choice. Remind us again, using easy to understand instructions is more conducive to non-professional users to accurately understand the key information.
This study found when the HIVST results were reactive more than 80% people choose to con rm by medical institution, when the results were nonreactive most people choose to regular retest, and when the results were invalid more than 80% people choose to test again. It is suggested that HIVST can help to nd more HIV infected people, to some extent [12,16].
If provided some help to HIV self-testers such as simple instructions and video tutorials, the HIVST results can be highly consistent with professional medical and health workers [17]. Studies showed the OMT HIVST accuracy is 97.0%[18], 92.5% [12], and 83.3% [19]. In this study, the HIVST accuracy of urine, OMT, and blood were 96.9% (255/263), 91.8% (56/61) and 100% (17/17). There was no factors affecting the HIVST accuracy by evaluation results analysis, the possible reasons were the HIV antibody content different between urine, OMT, and blood which due to the interfering substances of urine and OMT, and also may be because of the small sample size in this study.
Studies have shown that the urine of HIV-1 infections is unlikely contain infectious HIV-1, the risk of transmission of HIV-1 by urine is low to nonexistent [20]. And the urine sampling is non-invasive and painless which can improve testers to choose and acceptance. In this study, we found if HIVST was conducted again, 54.1% MSM tended to choose blood test reagents, 15.8% to choose blood and urine test reagents, 14.4% choose urine test reagents, and the reasons were people believes blood test is more accurate but urine test is more convenient. It is consistent with the study results of Witzel et al. [21] and Lippman et al. [22], which also showed that HIV blood test in MSM was more acceptable because of MSM believe it more accurate. However, the research of Marley  In conclusion, this study for the rst time to evaluated the understanding ability of key information in manual, the interpretation ability of simulation results, and the operation ability of HIVST, and the preference for urine, blood and OMT, when unprofessional MSMs used professional rapid test kits for HIVST lays a scienti c theoretical basis for further carring out HIVST in China.

Ethical requirement
The research was reviewed and approved by the ethics review committee of the center for STD and AIDS prevention and control of Chinese center for disease control and prevention (Project No. X171103483). All methods were performed in accordance with the relevant guidelines and regulations. All the documents during study are strictly con dential, and the identities of the participants in study were strictly con dential and free to opt out at all stages of the investigation.

Setting and population
This study was conducted in three HIV testing sites of MSM community organizations in Guiyang of Guizhou Province and Nanning of Guangxi Zhuang Autonomous Region. There are independent testing and consulting rooms to provide a private space for participants. Inclusion criterias for study participants were no mental illness or consciousness disorder, voluntary and willing to signed informed consent. Exclusion criterias were unwilling to sign informed consent, have been diagnosed with HIV or AIDS.
Investigators are experienced peer education volunteers and have received rapid test training. Investigators understood the purpose and signi cance of the study, familiar with the investigation process and questionnaire, mastered the process and operation of urine, OMT and blood HIV tests.

Questionnaire design and key information understanding evaluation
Questionnaires were designed in accordance with the instructions of rapid HIV detection reagents used in this study. Questionnaire including basic information of participants (age, nationality, marital status, educational level, residence, etc.) and the understanding of key information in instruction (important steps, results interpretation, considerations and limitations, in the face of different self-test results may take actions).
Participants ll their own basic information in the rst part, after viewing the instructions to answer and ll the questions about the key information in the second part. After completed the questionnaire, the investigators to check the questionnaire in time. If there are missing items or obvious errors in the questionnaire, the participants would be reminded to supplement and improve.

Results simulation pictures preparation and discriminant evaluation
Simulation pictures of test results such as weak reactivity, non-reactivity, and ineffectiveness (neither test line nor quality control line and only test line) were prepared in advance. The investigators showed the results simulation pictures to participants and asked them to interpret each simulation result, and recorded the interpretation results in the corresponding position.
Operation record table design and operation evaluation Selected 15 participants who had never received HIV test were divided into urine test group, OMT test group and blood test group, with 5 people in each group. Each group of every participants was given a correspond rapid HIV testing service package and completed HIV testing. From the moment of participants open the package, researchers begin to record in detail the all operation errors in preparation, sampling, testing, results interpretation, waste disposal and so on, without any help or guidance during the whole evaluation process. After the evaluation, asked participants in detail about their problems and confusion during the whole test process. Finally, according to the results of operation and conmunication to design the operation records including key steps in urine, OMT and blood HIVST.
After signed the informed consent, investigators rst provided the a urine testing service package (including a urine testing kit, instruction manual, disposable urine cup, disposable dropper, etc.). Under the premise of no guidance, all participants completed the urine HIV rapid testing by themselves. Afterwards, provided the OMT service package (including a OMT test kit, instruction manual, disposable oral swab, etc.) and blood testing service package (including a test kit, instruction manual, disposable blood needle, etc.) to some volunteer participants, and completed the HIV tests without any guidance. Investigators carefully observed the operation of each speci c step and lled in the operation record form, and lled the nal results interpretation by participants in the "Self-Test Results"section. At the same time, investigators lled their own results interpretation in the "Professional Results" section.

Data analysis
Removing the questionnaires with missing key information or non-conforming, used EpiData 3.1 and double entry method to establish the database, and used SAS 9.4 conducted data consistency tes. In this study, frequency and constituent ratio were used to represent enumeration data, while mean and standard deviation were used to represent measurement data.  Table 2