|
Polymerase chain reaction for the diagnosis of HIV infection in adults: a meta-analysis with recommendations for clinical practice and study design |
Owens D K, Holodniy M, Garber A M, Scott J, Sonnad S, Moses L, Kinosian B, Schwartz J S |
|
|
Authors' objectives To evaluate the sensitivity and specificity of polymerase chain reaction (PCR) assay for the diagnosis of human immunodeficiency virus (HIV) infection in adults. In addition, to develop recommendations for the design of future evaluative studies of PCR and use of PCR in patients with suspected HIV.
Searching Seventeen computer databases, including MEDLINNE, were searched from 1988 to 1991; MEDLINE was also searched from 1992 to 1994. The search strategy was given in the appendix of the paper. In addition, abstracts were identified from conference proceedings.
Study selection Study designs of evaluations included in the reviewThe studies had to include at least 10 participants.
Specific interventions included in the reviewStudies using DNA amplification by PCR on peripheral mononuclear blood cells were eligible for inclusion.
Reference standard test against which the new test was comparedNo inclusion criteria relating to the reference standard were specified. Details of the reference standards used in the included studies were not fully reported. The authors suggested enzyme immunoassay followed by confirmatory Western blot analysis, or viral culture as examples of reference standards used. They further suggested that serial testing or follow-up would be the most appropriate methods of establishing the absence of infection.
Participants included in the reviewStudies of participants aged at least 17 years were included.
Outcomes assessed in the reviewStudies reporting sufficient data for the calculation of sensitivity and specificity were eligible for inclusion.
How were decisions on the relevance of primary studies made?Two investigators independently examined all the titles, abstracts, and full articles identified in the search. Any disagreements were resolved by re-review and discussion.
Assessment of study quality Study validity was categorised according to a 4-point scale, based on a developed assessment framework (see Other Publications of Related Interest nos.1-3). This framework addressed: appropriateness of patient spectrum; description of selection criteria; reporting of details of the index test; use of an appropriate reference standard; verification bias (whether the same reference standard was used and whether all patients received it); and review bias (blinding of investigators to the results of other tests and/or clinical information when interpreting test results). Two independent investigators, who were blinded to the authors, research institution, journal, study title and results, assessed validity.
Data extraction Two investigators independently abstracted the data from each study. Any disagreements were resolved by re-review and discussion.
Methods of synthesis How were the studies combined?The summary receiver operating characteristic (sROC) curves were estimated using a logistic transformation of sensitivity and specificity. The sROC curves were symmetrical (no threshold effect), therefore, the Mantel-Haenszel method was used to estimate the common odds ratio. sROC curves based on both the upper (indeterminate PCR results excluded) and lower (indeterminate PCR results counted as false-positive or false- negative results) estimates of sensitivity and specificity were reported. The point of maximum joint sensitivity and specificity was used to compare the sROC curves.
How were differences between studies investigated?A subgroup analysis was carried out on studies published as abstracts only, study quality score, recent studies, and specific aspects of study design.
Results of the review Ninety-six studies with 14,668 participants (5,739 with HIV infection and 8,929 without) were included.
When indeterminate PCR results were excluded, the sensitivity ranged from 10 to 100% and the specificity from 40 to 100%. In studies that were attributed the top two rating categories, the sensitivity ranged from 83 to 100% and the specificity from 95 to 100%.
For all studies, the upper estimate (indeterminate PCR results excluded) of the maximum joint sensitivity and specificity was 98% and the lower estimate (indeterminate PCR results counted as false-positive or false-negative results) was 97%. The corresponding log odds ratios (+/- SE) were 7.93 (+/- 0.330) and 6.96 (+/- 0.195), respectively. The exclusion of studies where sensitivity and specificity were derived from the number of samples rather than the number of patients, or studies where heparin was used to preserve the blood samples, did not significantly affect the results.
Subgroup analysis: the upper estimate of maximum joint sensitivity and specificity, based on studies that received the top three score categories, did not differ significantly from the estimate based on studies with the lowest quality score. However, the lower-bound estimate of joint sensitivity and specificity was significantly lower in studies with better scores. Studies published during or after 1991 gave lower estimates of accuracy than those published before 1991. Studies published as abstracts only reported lower values of sensitivity and specificity than did those published as full articles.
Post-test probability: at the maximum joint sensitivity and specificity, the positive predictive value of PCR ranged from 34 to 85% as the prevalence of HIV infection increased from 1 to 10%.
Authors' conclusions The PCR assay was not sufficiently accurate for the diagnosis of HIV infection without confirmation. Currently, the interpretation of PCR test results should be combined with careful consideration of the clinical circumstances, and with the use of confirmatory tests and clinical follow-up whenever possible. It is anticipated that technical advances may eventually improve the performance of the PCR assay.
CRD commentary The review addressed a clear question and the inclusion criteria were defined. The analysis was appropriate and comprehensive, and study quality was assessed and its impact on results investigated. However, the absence of inclusion criteria defining the reference standard or clear reporting of the reference standards used is a significant problem. The search strategy was very limited and was restricted to published, English language studies. It therefore seems likely that relevant data may have been omitted. No attempt to assess publication bias was reported. The reporting of the characteristics of the included studies was poor, making it difficult to assess the generalisability of the results.
The authors' conclusions follow from the results presented. However, the review should be interpreted with caution given the limitations described.
Implications of the review for practice and research Practice: The authors stated that the use of PCR for the diagnosis of HIV in adults should be limited to situations in which antibody tests are known to be insufficient.
Research: The authors stated that future studies of PCR should be sufficiently large, and should use adequate reference standard tests and standardised methods for PCR. Specimens should be evaluated by persons blinded to clinical status and to the results of other diagnostic tests for HIV infection.
Bibliographic details Owens D K, Holodniy M, Garber A M, Scott J, Sonnad S, Moses L, Kinosian B, Schwartz J S. Polymerase chain reaction for the diagnosis of HIV infection in adults: a meta-analysis with recommendations for clinical practice and study design. Annals of Internal Medicine 1996; 124(9): 803-815 Other publications of related interest 1. Hoffman RM, Kent DL, Deyo RA. Diagnostic accuracy and clinical utility for lumbar radiculopathy: a meta analysis. Spine 1991;16:623-8.
2. Kent DL, Haynor DR, Larson EB, Deyo RA. Diagnosis of lumbar spinal stenosis in adults: a metaanalysis of the accuracy of CT, MR, and myelography. AJR Am J Roentgenol 1992;158:1135-44. 3. Kent DL, Larson EB. Disease, level of impact, and quality of resarch methods. Three dimensions of clinical efficacy assessment applied to magnetic resonance imaging. Invest Radiol 1992;27:245- 54.
Indexing Status Subject indexing assigned by NLM MeSH Adult; HIV Infections /diagnosis; Humans; Polymerase Chain Reaction; ROC Curve; Reproducibility of Results; Research Design; Sensitivity and Specificity AccessionNumber 11996008203 Date bibliographic record published 30/04/2004 Date abstract record published 30/04/2004 Record Status This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn. |
|
|
|