Quality Assessment Tables

Cynthia Feltner; Catherine Grodensky; Charles Ebel; Jennifer Cook Middleton; Russell P Harris; Mahima Ashok; Daniel E Jonas

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Feltner C, Grodensky C, Ebel C, et al. Serological Screening for Genital Herpes: An Evidence Review for the U.S. Preventive Services Task Force [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2016 Dec. (Evidence Syntheses, No. 149.)

Cover of Serological Screening for Genital Herpes

Serological Screening for Genital Herpes: An Evidence Review for the U.S. Preventive Services Task Force [Internet].

Show details

Contents

< Prev Next >

Appendix DQuality Assessment Tables

Appendix D Table 1Quality Ratings of Studies Assessing the Accuracy of Serologic Screening Tests for HSV-2 (Key Question 2)

First Author, Year	Was the cutpoint used to determine test positivity adequately described (or referenced)?	Were population selection criteria clearly described?	Did the whole or a random selection of the participants receive the Western blot?	Did all participants receive the Western blot regardless of serologic screening test results?	Were the serologic test results and Western blot results interpreted independently?
Lingappa, 2010⁷³	Yes	Yes	Yes	Yes	Yes
Mark, 2007⁶⁵	Yes	Yes	Yes	Yes	NR/CND
Ng’ayo, 2010⁷⁴	Yes	Yes	Yes	Yes	NR/CND
Delany-Moretlwe, 2009⁷⁵	Yes	Yes	Yes (random selection)	Yes	NR/CND
Summerton, 2007⁹⁵	Yes	NR/CND	See comments	No	NR/CND
Ashley-Morrow, 2004⁶⁶	Yes	Yes	NR/CND	Yes	Yes
Mujugira, 2011⁶⁷	Yes	Yes	Yes	Yes	Yes
Smith, 2009⁶⁸	Yes	Yes	Yes	Yes	NR/CND
Golden, 2005⁶⁹	Yes	Yes	Yes	No	NR/CND
Morrow, 2005⁷⁰	Yes	Yes	NR/CND	No	NR/CND
Hogrefe, 2002⁷¹	Yes	Yes	Yes	Yes	NR/CND
Gamiel, 2008⁹⁶	Yes	No	Yes	Yes	NR/CND
Van Dyck, 2004⁷²	Yes	No	Yes	Yes	NR/CND
Ashley, 1998⁹⁷	No	No	No	Yes	NR/CND

First Author, Year	What was the overall attrition?	Were withdrawals from the study explained (post-enrollment)?	Were methods for calculating accuracy clearly reported and valid?	Did the study have high attrition raising concern for bias?	What was the method used to handle missing data?	Quality	Comments
Lingappa, 2010⁷³	5%	Yes	Yes	No	Excluded	Good	5% (N=26) of samples had equivocal WB results and were excluded from the analyses. The characteristics of the subset of participants included in this analysis were not described (only those of the overall community cross-sectional sample, N=1124). There was no description of whether participants had current or previous symptoms consistent with genital herpes.
Mark, 2007⁶⁵	11%	Yes	Yes	No	Excluded	Good	NA
Ng’ayo, 2010⁷⁴	≥6% (see comments)	Yes	Yes	Unclear	Excluded	Fair	Characteristics of population not described (included prior symptoms of genital herpes); all equivocal and indeterminate results (on both WB and serologic screening test) were excluded from sensitivity and specificity calculations. For higher cutoff values on the Focus test, the number of equivocal values was high (approximately 40% of the sample tested).
Delany-Moretlwe, 2009⁷⁵	Unclear	NA	Yes	No	NA	Fair	A random sample of results from the larger sample (N=210) was compared with WB; the results were used to extrapolate sensitivity/specificity in the full sample. Handling of Indeterminate or equivocal test results was not reported. Results for subgroups of participants (by age and HIV status) were given but no measure of variance (confidence interval) was reported for the subgroups.
Summerton, 2007⁹⁵	1%	Yes	Yes	No	Excluded	Poor	Specificity outcome was not eligible due to sampling strategy. All participants who had a positive result on at least 1 of 3 serologic screening tests had the WB; participants who had a negative result on the 3 serologic tests were excluded.
Ashley-Morrow, 2004⁶⁶	See comments	Yes	Yes	No	NA	Fair	Samples from some sites (Barcelona and Hanoi) were not considered due to technical issues. 20 samples were excluded due to equivocal results (2.9%). Subset of samples were compared to the WB and results were used to estimate the sensitivity/specificity for the overall sample.
Mujugira, 2011⁶⁷	4%	Yes	Yes	No	Excluded	Good	Unequivocal test results (4%) excluded from sensitivity/specificity calculations.
Smith, 2009⁶⁸	2%	NA	Yes	No	Excluded	Fair	Blinding of outcome assessors is not reported (but the tests were conducted at different sites); does not appear that data were missing but equivocal results were excluded.
Golden, 2005⁶⁹	5%	NR/CND	Yes	No	Excluded	Fair	Unclear if test results were interpreted blindly; excluded atypical WB results.
Morrow, 2005⁷⁰	8%	NR/CND	Yes	No	Excluded	Fair	Testing was not performed on the whole sample; only the MSM sample was reported to be randomly selected (the other sample was not reported). All participants did not receive WB; assuming blinded.
Hogrefe, 2002⁷¹	2%	NR/CND	Yes	No	See comments	Fair	WB atypical tests were excluded; indeterminate HerpeSelect serologic test results were considered positive.
Gamiel, 2008⁹⁶	NR	NA	NR/CND	NA	NR	Poor	Methods for calculating sensitivity/specificity are not reported, specifically how indeterminate values were handled. Sample size for the Biokit HSV-2 Rapid Test analysis for HIV-negative subgroup is not reported.
Van Dyck, 2004⁷²	Unclear	NA	Yes	NR/CND	NR	Fair	Sensitivity was estimated by taking a random sample of serologic test results that were concordant (positives and negatives) and all those that were discordant compared with the monclonal antibody-blocking enzyme immunoassay and comparing those with the WB. The handling of indeterminate tests is unclear; however, it appears that a positive test was defined as ≥1.1, and lower results were considered negative.
Ashley, 1998⁹⁷	2%	NA	Yes	No	Excluded	Poor	Characteristics of study sample are not reported. Risk of spectrum bias; samples were chosen based on known, clear profiles to HSV-1 and HSV-2.

: Abbreviations: CND=cannot determine; N=number; NA=not applicable; NR=not reported; MSM=men who have sex with men; WB=Western blot.

Appendix D Table 2Quality Ratings of Studies Assessing the Harms of Serologic Screening for HSV-2 (Key Question 3)

First Author, Year	Were eligibility criteria clearly described?	Were subjects^* representative of the overall source population?	Were criteria used to assess prior symptoms clearly described?	What was the overall attrition?	Did the study have high attrition raising concern for bias?	Were outcomes prespecified/defined and adequately described?	Were outcome measures valid and reliable?	Quality	Comments
Smith, 2000⁹⁸	Partially	NR/CND	No	46%	Yes	Yes	Yes	Poor	High risk of selection bias and high attrition (with persons having significant anxiety less likely to follow up). Study population was recruited from persons presenting to sexual health clinics in Australia. How authors determined symptom status was not described.
Edlow, 2012⁹⁹	No	NR/CND	No	NR	NR/CND	Yes	Unclear	Poor	This is an abstract that has limited description of methods, including no description of eligibility criteria. Overall, there is a high risk of selection bias. Personal communication from the author indicated that “all comers” were enrolled. The author states that participants were not explicitly treatment-seeking or symptomatic; however, the percent with no prior or current symptoms is unknown. Validity of the GHQ-12 to assess harms (e.g., whether it is sensitive enough) in this context is uncertain.
Mark, 2008⁷⁷	Partially (see comments)	NR/CND	No	72%	Yes	Yes	Yes	Poor	High risk of selection bias, high attrition, and no control group. Very small sample with just 3 WB confirmed positives completing the followup. Eligibility criteria are not clear about determination of history of genital sores or genital herpes or whether participants were required to be asymptomatic. Given the heavy reliance on flyers and ads for recruitment, population is more likely a group with possible reasons to want testing.
Rosenthal, 2006⁷⁹	No	NR/CND	No	19%	Yes	Yes	Yes	Fair for HRQOL outcomes; poor for all other outcomes	High risk of selection bias, very high attrition, no concurrent control group that was not screened. Participants required to have no known history of genital herpes; criteria used to determine symptoms was not described. We rated herpes-related QOL data as fair quality; the lack of a control group for this outcome is less concerning since the questions are specific to having a genital herpes diagnosis.
Melville, 2003⁷⁸	Yes	No	No	67% of those invited participated (24/36)	Yes	Yes	Yes	Fair	High risk of selection bias; participants were selected from various sites using different recruitment procedures. It is unclear how many were eligible at each site. Authors note 67% of those invited agreed to participate. Authors used predefined semistructured interviews to elicit psychosocial outcomes related to serologic testing. The questionnaire is not shown. Only themes reported by ≥3 participants were reported, so less common outcomes and potentially serious outcomes (suicidality) may have occurred in as many as 2 participants without being reported.
Hallfors, 2015⁴⁵	Yes	No	No	Unclear	NR/CND	No	Unclear	Poor	Participants were orphans selected from 26 primary schools in Nyanza Province, Kenya. It is unclear whether participants were asked about history of signs/symptoms of genital herpes. Proportion of youth who declined to participate was not reported. Data on the psychosocial response at disclosure appears to have been collected on all 28 participants who tested positive. Outcome measures are not described; research staff and interviewers coded participant and caregiver responses to disclosure. Results were not based on patient-reported (or caregiver reported) measures of psychosocial harms.
Richards, 2007⁸⁸	Yes	NR/CND	No	Of those testing HSV-2 positive (N=87), 89% completed followup	Yes	Yes	Mixed	Poor	High risk of selection bias; unclear if subjects who agreed to participate are similar to the overall source population. Recruitment was not based on the presence or absence of prior symptoms; half of participants were HSV-2 positive or had a prior diagnosis of genital herpes. Criteria for determining prior diagnosis of genital herpes are not described. Of those contacted via letter (N=5703), 17% responded and agreed to be contacted. Of those who agreed to be contacted (N=955), 36% agreed to enroll and have HSV-2 testing, 29% declined to participate, 33% could not be contacted, and 2% were ineligible. Of those who tested HSV-2 positive (N=87), 89% completed followup. Many outcomes were general QOL or mood state; it is unclear if these are valid measures of the harms associated with HSV-2 screening.

*: Are they generally asymptomatic persons with no prior history of genital herpes recruited from primary care settings? Is the sample that participated similar to the overall source population?
: Abbreviations: CND=cannot determine; N=number; NA=not applicable; NR=not reported; QOL=quality of life; WB=Western Blot.

Appendix D Table 3Quality Ratings of Studies of Antiviral Medications on HSV-2 Viral Shedding, Symptomatic Episodes, and Transmission (Key Questions 4 and 5)

First Author, Year	Was randomization adequate?	Was allocation concealment adequate?	Are baseline characteristics similar between groups?	What was the overall attrition?	What was the differential attrition?
Corey, 2004⁷⁶ Kim, 2008⁸³	Yes	Yes	Yes	22%	2%
Mujugira⁸⁴	Yes	NR/CND	Yes	66%	2%
Sperling, 2008⁸¹	Yes	NR/CND	Yes	29%	1.3%
Leone, 2007⁸²	Yes	Yes	Yes	23%	NR

: Abbreviations: CND=cannot determine; NR=not reported.

First Author, Year	Did the study have high differential attrition (>10%) or overall high attrition (generally 20%) raising concern for bias?	Did the study have crossovers or contamination raising concern for bias?	Were outcome measures valid and reliable?	Were patients masked?	Were providers masked?	Were outcome assessors masked?	Was the duration of followup adequate to assess the outcome?
Corey, 2004⁷⁶ Kim, 2008⁸³	No	No	Yes	Yes	NR/CND	Yes	Yes
Mujugira⁸⁴	Yes	NR/CND	Yes	Yes	Yes	NR/CND	Yes
Sperling, 2008⁸¹	Yes	No	Yes	Yes	Yes	NR/CND	Yes
Leone, 2007⁸²	Yes	NR/CND	Yes	Yes	Yes	NR/CND	Yes

: Abbreviations: CND=cannot determine; NR=not reported.

First Author, Year	What was the method used to handle missing data?	Did the study use acceptable statistical methods? ITT vs. per protocol; adjustment for factors?	Was compliance to study medication adequate?	Quality	Comments
Corey, 2004⁷⁶ Kim, 2008⁸³	Data for subjects who did not reach an end point were censored as event-free periods ending on the last day that the absence of the end point was confirmed.	Yes	Yes	Fair	More couples randomized to placebo withdrew; per authors withdrawal occurred because more source partners had frequent symptoms. Missing data were censored (as event-free periods). However, differential attrition was relatively low (2%).
Mujugira⁸⁴	Modeling was used to impute some data.	Yes	Yes	Fair	Overall attrition is high; analysis accounted for some of the missing data.
Sperling, 2008⁸¹	Missing data were excluded.	Yes	NR/CND	Fair	This is a crossover RCT. There was an overall high rate of attrition (29 %); differential attrition was low.
Leone, 2007⁸²	Unclear; modeling was used to estimate differences between groups and likely some data was imputed. However, participants who provided no swabs were excluded from analysis.	Yes	Yes	Fair for KQ 4 and 5 outcomes	This is a cross-over RCT; overall attrition is high (23%). Handling of missing data is unclear for some outcomes.

: Abbreviations: CND=cannot determine; NR=not reported; RCT=randomized, controlled trial.

Appendix D Table 4Quality Ratings of Studies of Assessing Harms Of Preventive Interventions (Key Question 6)

First Author, Year	Were harms prespecified and defined?	Were ascertainment techniques for harms adequately described?	Were ascertainment techniques for harms equal, valid, and reliable?	Was duration of followup adequate for harms assessment?	Quality	Comments
Sperling, 2008⁸¹	No	No	NR/CND	Yes	Fair	Adverse events were assessed at every visit after discussion with the subject and review of the subject’s diary. Harms do not appear to have been prespecified.
Leone, 2007⁸²	No	No	NR/CND	Yes	Poor	Harms are reported but not prespecified for well-defined; harms are only reported for the overall sample and not the subgroup of participants with no prior history of genital herpes.

: Abbreviations: CND=cannot determine; NR=not reported.

Bookshelf ID: NBK409118

Contents

< Prev Next >

PubReader
Print View
Cite this Page
Feltner C, Grodensky C, Ebel C, et al. Serological Screening for Genital Herpes: An Evidence Review for the U.S. Preventive Services Task Force [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2016 Dec. (Evidence Syntheses, No. 149.) Appendix D, Quality Assessment Tables.
PDF version of this title (968K)

Other titles in these collections

Recent Activity

Clear Turn Off Turn On

Quality Assessment Tables - Serological Screening for Genital Herpes
Quality Assessment Tables - Serological Screening for Genital Herpes

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

Bookshelf