From settings where recommended infection control measures are applied, only a few cases of HIV-transmission from patient-to-patient and from health care workers to their patients have been reported[1-5]. Chant et al.  reported HIV-transmission from one patient to four other patients undergoing minor surgery on the same day carried out by a (HIV-negative) general surgeon. The case, in which an HIV-positive dentist transmitted the infection to six of his patients, by an unknown route, has been widely quoted[4,6-10]. In that case the transmissions were rendered probable by a combination of epidemiological investigations and molecular biological analyses.
When a women (patient GP) with common variable immunodeficiency, who regularly received intravenous immunglobulin IgG treatment, and with no known risk factors for HIV-infection, surprisingly tested HIV-antibody positive, we decided to investigate the possibility of horizontal nosocomial infection.
Among patients receiving intravenous treatments at the out-patient facility at the same day as GP, we identified a patient (FDL) with HIV that was highly related to the virus found in GP. We describe the epidemiological investigation, the results of the genotypic analyses and a possible way of transmission.
Patients and methods
Patient GP most likely contracted the HIV-infection between October 1994 and February 1996 (see below). During this period GP received intravenous immunglobulin on 17 occasions in our outpatient facility. The immunoglobulin batches given to GP were identified, and found not to contain HIV. All of the patients with humoral immunodeficiency treated with immunoglobulin substitution in the out-patient clinic were identified and tested for HIV-antibodies and antigen; all tested negative except GP. Hence, transmission through contaminated immunoglobulin was ruled out.
About one thousand patients are treated at the clinic, the majority (approximately 700) of whom are infected with HIV. In addition to regular control visits, minor interventions e.g. intravenous chemotherapy and blood transfusion are performed at the clinic. From one to six patients have these procedures performed daily.
Through appointment records we identified the patients, who had received intravenous treatments (e.g. intravenous medication or blood transfusion) in the outpatient facility on the same day as GP during the 16-month period (October 1994-February 1996). Thirteen patients had done so, and were HIV-infected. These patients and one patient with an ill-defined way of HIV transmission (patient AM) were included in the analyses.
An interview with GP disclosed that, she had never shared a room with any patients while receiving intravenous therapy.
In 1992, patient GP, a woman aged 58 years at that time, was referred to our ward under the diagnosis common variable immunodeficiency. The diagnosis was based on accumulated episodes of pneumonia, eight in less than 1 year, and the finding of hypogammaglobulinaemia. Immunoglobulin substitution therapy was initiated. During 1996 the patient experienced malaise and had a 10kg loss of weight. Malignancy was suspected, but could not be confirmed. In March 1997 the patient had HIV-antibody and antigen tests performed, and was found to be HIV positive. No risk factors for HIV infection could be identified. The patient had not had sexual relationships since 1985, had no records of intravenous drug abuse and had never received blood transfusions. HIV RNA analyses (Amplicor HIV Monitor, Roche Diagnostics Systems, Inc., Branchburg, NJ, USA) on stored serum samples revealed, that GP was HIV RNA negative in March 1995 and positive in February 1996. It was therefore assumed that the patient had contracted the HIV infection between October 1994 and February 1996. A 6-month period prior to HIV RNA positivity was chosen as a conservative estimate. Longitudinal CD4 cell counts, viral loads, plasma IgG and IgM levels are shown in Fig. 1.
Possible source patients
The clinical data on the possible source patients are shown in Table 1, including the duration of known HIV-seropositivity, AIDS-defining events prior to/at the time of suspected transmission, and the number of times the patients received intravenous therapy on the same day as GP. The CD4 cell counts and serum viraemia levels are included when available.
FDL first tested HIV-positive in 1988; he was HIV-antibody negative in 1987. He was diagnosed with AIDS in 1994 on the basis of Pneumocystis carinii pneumonia. In January 1995 FDL developed Cytomegalovirus chorioretinitis. Outpatient treatment with foscarnet via a Port-a-cath (Pharmacia, Copenhagen, Denmark) was initiated and kept up until his death in September 1995. Due to psychiatric illness home treatment with foscarnet was not feasible. FDL was treated on the same day as GP on five occasions between April and September 1995. Stored serum samples from patient FDL from this period were retrospectively analysed for HIV RNA content (Table 1).
Twenty local HIV-positive patients were included as a control group. Furthermore reference sequences from various subtypes (A-H), obtained from the Los Alamos database, were included in the phylogenetic analyses.
Stored serum samples from the 14 possible source patients, obtained around the time of suspected transmission, were identified. We performed reverse transcriptase (RT)-polymerase chain reaction (PCR) and subsequent sequencing of the C2-V3-C3 region of the env region, as well as amplification and sequencing of the p17-encoding region of gag on a subset of the samples. For samples from GP and FDL we also performed sequencing of the RT-region. Samples (peripheral blood mononuclear cells and plasma) from GP were analysed after the analyses of the possible index patients and was confirmed on a second plasma sample.
For the env amplification we used the primer JA 167:5′-TAT C(C/T)T TTG AGC CAA TTC C(C/T)A TAC A-3′?and JA 170:5′-GTG AGT TAT T(A/G)C A(A/G)T AGA AAA ATT C-3′?for the first PCR, and primers JA 168:5′-ACA ATG (C/T)AC ACA TGG AAT TA(A/G) GCC A-3′ and JA 169:5′-AGA AAA ATT C(C/T)C CTC (C/T)AC AAT TAA A-3′?for the second PCR, yielding a 410 base pair (bp) long amplification.
For the RT-PCR of the p17-encoding region we used primers JA 152:5′- ATC TCT AGC AGT GGC GGC CGA ACA G-3′ and JA 155:5′-CTG ATA ATG CTG AAA ACA TGG GTA T-3′?for the first PCR and primers JA 153:5′- CTC TCG ACG CAG GAC TCG GCT TGC T-3′ and JA 154:5′- CCC ATG CAT TCA AAG TTC TAG GTG A-3′?for the nested PCR. A fragment of 558 bp was amplificated.
For the RT-region we used a nested PCR with primers L3 (5′-GAC CAG AGC CAA CAG C-3′) and L4 (5′-ATC ACT AGC CAT TGC TCT CCA-3′) for the first PCR and primers A (5′-TTC CCA TTA GTC CTA TT-3′) and L31 (5′-CCA GCT GTC TTT TTC TGG CAG CAC TAT-3′) for the nested PCR.
The PCR products were analysed on 1.5% agarose gels and visualized by ethidium bromide staining. Prior to sequencing, the PCR products were purified using Qiaquick PCR purification kit (Qiagen GmbH, Germany) according to the manufacturers instructions.
The direct sequencing procedures were performed with the primers used in the second PCRs using the Dye Terminator Cycle Sequencing Kit (Perkin Elmer) and analysed on the ABI PRISM 377 Sequencer.
Analyses of sequences
All multiple nucleotide alignments were performed with CLUSTAL W (v 1.75). The aligned sequences correspond to the following positions in HXB2: env position 7002-7323, gag 789-1137 and RT to 2594-3274. Phylogenetic analyses were performed using the neighbor-joining method  as implemented in CLUSTAL W. Trees were plotted using NJPlot (kindly provided by Manoo Guy of the University of Lyon, France) and DRAWTREE (Phylip). The trees were bootstrapped 10000 rounds and the significances of the nodes shown are given in percentage. Distances were corrected for multiple substitutions ad modum Kimura.
The env sequences from GP obtained from peripheral blood mononuclear cells and plasma were very nearly identical (0.3% difference). Hence, only the sequences obtained from plasma-derived HIV RNA were included in the subsequent analyses.
We were able to obtain env sequences from eleven of the fourteen possible source patients. The sequences from GP and nine of the possible source patients clustered with the subtype B reference sequence, whereas two patients harboured subtype D sequences (Fig. 2). Both patients with subtype D had lived in Tanzania, and were though to have contracted the HIV-infection there (Fig. 2).
The phylogenetic tree analyses showed clustering between the sequences obtained from GP and FDL (Figs 2 and 3).
The C2-V3-C3 sequence obtained from GP was almost similar to the one from patient FDL. The pairwise nucleotide distance between GP and FDL was 1.9%. In contrast the pairwise distance between the sequences recovered from GP and the other ten patients ranged between 8.9 and 22%. The distance between GP and the local controls (LC1-LC20) ranged between 7.9 and 13.9%. The former ranking higher due to the inclusion of two patients infected with subtype D, whereas all the local controls were infected with subtype B HIV. The bootstrap linking GP to the same cluster as FDL was 99.9%.
In addition to FDL, three of the other patients had gag analyses performed, including one in whom we were unable to amplify the env region. This analysis confirmed the similarity of the sequences from GP with those from FDL, with the pairwise nucleotide distance between GP and FDL being 0.9%, as opposed to 6.6, 7.2 and 12.6% between GP and the other three patients, respectively. The bootstrap value for the node between GP and FDL was 100% (Fig. 4).
The RT amino acid sequences from GP and FDL differed by 0.9% (2 of 227). GP harboured HIV RNA with the foscarnet mutation W88S. The latest plasma sample available from FDL was collected in March 1995. At this time FDL had received foscarnet for 6 weeks. The plasma RNA did not contain the W88S mutation. Patient FDL had been treated on the same day as GP on five occasions. HIV RNA analysis of stored serum samples from FDL showed high-grade viraemia (>105 copies/ml) during this period (Table 1).
Changes in IgG and IgM
Interestingly, GP‚s plasma IgG rose transiently after the acquisition of the HIV-infection (Fig. 1), leading to a reduction in the dosage of the substitution therapy. The plasma IgG level dropped again, coinciding with highly active antiretroviral therapy (HAART)-induced viral suppression. Parallel changes were seen in plasma IgM (Fig. 1), whereas no changes in the IgA level could be detected (data not shown).
Way of transmission
The precise timing of the transmission could not be identified, but could be narrowed down to the period April to September 1995. During this period patients GP and FDL received intravenous treatment on the same day on five occasions.
Examination of the routines in connection with intravenous treatments of patients in the outpatient clinic did not disclose routine procedures, which could lead to a breach in the infection control measures. However, it was found that health care workers occasionally under stress could abandon routine procedures in the following way. When an intravenous access is established, before and after the administration of medicine or blood the drip is flushed with sterile saline. We used to utilize 50-ml bottles with permeable membranes. These bottles were changed daily. After use for an individual patient the needles were discarded. It might be that a health care worker drew additional saline from a bottle and forgot to dispose it after use. Hence, a health care worker flushing FPL‚s Port-a-cath might have needed more saline, and have drawn additional saline without changing the syringe. In this way the saline could have been microscopically contaminated with blood, and HIV. If saline from the same bottle was used for flushing GP‚s drip, this could have led to horizontal transmission. The procedure has since been changed. Small ampoules with breakable necks are now used for saline supplement and disposed after each individual use.
Only a few cases of nosocomial patient-to-patient HIV-transmission have been reported[1-3,15]. The basis for the detection of nosocomial transmissions is a high degree of attention towards unexpected cases of HIV and careful epidemiological investigations. The ability to perform viral genetic sequence analyses is increasingly being used to validate the results of epidemiological investigations. A requirement for such genetic studies is the existence of viral genetic variation. The greater the variation, the greater the power of such methods to distinguish different strains of the virus. HIV has a high mutation rate, and hence, the genetic make-up differs between HIV-infected individuals.
Using sequences from the env, gag and the RT region of pol we were able to identify a person, FDL, with HIV RNA that was highly related to the virus harboured by GP. We used the C2-V3-C3 region of the env gene for analyses of the genotypic relatedness between patient GP and the patients identified as possible source patients and local controls.
The env sequence obtained from FDL was highly related to the one obtained from GP. The nucleotide sequence distance between GP and FDL was 1.9%. These values were substantially lower than the values obtained when comparing GP‚s sequences with those of the other possible source patients and the local controls. The nucleotide sequence distances between GP and FDL were similar to those previously found in epidemiologically linked cases[6,15-18].
The findings from the env gene were confirmed by the analyses of the gag gene. This analysis also found a clustering of the sequence from GP with the sequence from FDL. We found a distance of 0.9%, which is similar to the 2.2% difference found in a case of mother-to-child transmission.
As mentioned above, the env region has previously been used in studies of epidemiologically linked infections[4,5,16,17,20-22]. It has, however, been claimed that the env region, and especially the V3 loop, might not be optimal for detecting linkage due to convergent evolution. The V3 region contains the principal neutralization domain, and it has been suggested that analyses of the gag region might be more informative. Leitner et al.  found trees derived from V3 sequences to be more accurate than those derived from p17 data. Including sequence data from both the env and gag region yielded still more accurate estimates. Other studies have confirmed that the p17 region is epidemiologically informative[19,22,25,26].
Further support for a link between the virus from patients FDL and GP stem from the finding of the W88S foscarnet-related mutation in the reverse transcription region in HIV RNA derived from GP. This mutation was first described by Mellors et al., and causes a three- to four-fold reduction in susceptibility of HIV to foscarnet. Patient FDL had received foscarnet treatment, whereas none of the other possible source patients had. We did not find the W88S mutation in the HIV RNA from FDL. However, the latest sample available was collected in March 1995. At this time FDL had only received foscarnet for 6 weeks, and the mutation had probably not been selected for. In the study by Mellors et al.  the mutation was only detected in patients, who had received foscarnet. Hence, we presume that FDL developed the mutation after the plasma sample was collected and prior to the HIV transmission. To our knowledge this is the first report of transmission of foscarnet-resistant HIV.
Leitner et al.  have shown, that most of the methods for phylogenetic construction (Fitch-Margoliash, neighbor-joining, maximum-parsimony and maximum-likelihood) perform well. Others [28-30] have documented the usefulness of the neighbour-joining method for analyses of epidemiological linkage.
Both vertical and sexual transmission rates have been shown to correlate with viral load[31-34]. The higher the viral load the greater the risk of transmission. At the time of HIV-transmission patient FDL had high-grade viraemia, with serum viral load >105copies/ml. The inoculum needed to establish HIV infection is unknown. Apetrei et al.  have estimated the residual volume in syringes used in children with high-grade viraemia, and found that the residual volume could contain virions equivalent to 200 HIV RNA copies.
Patient GP experienced increases in serum IgG and IgM levels in conjunction with the HIV infection (Fig. 1). Wright et al.  described a patient with common variable immundeficiency, who had normalization of serum immunoglobulin (IgG, IgM and IgA) synthesis, development of specific immune responsiveness, and an improved clinical status, after being infected with HIV. Cases with increase in only the IgM, or IgM and IgG subclasses have also been described[37,38]. In the case of GP, the increase in IgG and IgM levels were of short duration. The decreases in serum levels coincided with antiretroviral-induced suppression of the viral replication, although chronologically also with a reduction in the immunoglobulin dosing (from 40 to 20 g/month from July 1997) (Fig. 1). To our knowledge this is the first report of HAART-induced reduction of immunoglobulins back to pathological levels in a common variable immunodeficiency-patient with HIV-induced increases in immunoglobulin levels. The basic mechanism for the defect in B-cell maturation or function among patients with common variable immunodeficiency is unresolved[36,39,40]. Hypergammaglobulin is a common finding among HIV-infected patients. The polyclonal immunglobulin stimulation induced by HIV antigens, or HIV-induced changes in the cytokine profile, might lead to (partial) restoration of the B-cell deficiency in patients with common variable immunodeficiency, as exemplified by the description by Wright et al., and to a certain extent also by the present case, but not the case reported by Gutierrez and Kirkpatrick.
As is the case for the other reports of patient-to-patient [1-3,15] and health care worker-to-patient  HIV transmissions, the exact way of transmission in the present case has not been established. However, a careful examination of the procedures in conjunction with handling of intravenous therapy disclosed a step, which, under stressful circumstances, could lead to a failure in the infection control measures. Contamination of multidose vials have previously been implicated in cases with nosocomial infection[43-45]. In these case the transmissions were strongly linked to the use of multidose vials and other routes of transmission were ruled out. In addition to these cases other reports have pointed in the same direction, although without substantial proof[46,47]. Hence, even in settings with a high degree of awareness regarding infection-control measures mistakes occur and the use of multidose vials seems to increase this risk. Recently, Widell et al. recommended that the use of multidose vials be restricted, and the present study lends support to that recommendation.
In conclusion, the present study confirms the ability of combined epidemiological and genotypic analyses to trace HIV transmission. By comparing env and gag sequences we could detect a very close similarity between HIV RNA from the index patient and HIV from one of the possible sources identified through the epidemiological investigation. GP harboured HIV RNA with the foscarnet-induced W88S mutation, further lending support to HIV RNA from the foscarnet-treated FDL being the source of the infection. Even though the literature shows that the risk of HIV transmission in the health care setting is low in developed countries[48,49], the present report as well as others[1,4,43-47], underlines the importance of a constant survey of the infection-control precautions in order to detect potential breaches.
1. Chant K, Lowe D, Rubin G, et al
. Patient-to-patient transmission of HIV in private surgical consulting rooms. Lancet
2. Blank S, Simmonds RJ, Weisfuse I, Rudnick J, Chiasson MA, Thomas P. Possible nosocomial transmission of HIV. Lancet
3. Nielsen H, Rosth¢j S, Machuca R, Nielsen C, Smith E. Nosocomial child-to-child transmission of HIV. Lancet
4. Ou CY, Ciesielski CA, Myers G, et al
. Molecular epidemiology of HIV transmission in a dental practice. Science
5. Blanchard A, Ferris S, Chamaret S, Guétard D, Montagnier L. Molecular evidence for nosocomial transmission of human immunodeficiency virus from a surgeon to one of his patients. J Virol
6. Ciesielski C, Marianos D, Ou CY, et al
. Transmission of human immunodeficiency virus in a dental practice. Ann Intern Med
7. Ciesielski CA, Marianos DW, Schochetman G, Witte JJ, Jaffe HW. The 1990 Florida dental investigation. The press and the science. Ann Intern Med
8. Myers G. Molecular investigation of HIV transmission. Ann Intern Med
9. Korber B, Myers G. Signature pattern analysis: a method for assessing viral sequence relatedness. AIDS Res Hum Retroviruses
10. Hillis DM, Huelsenbeck JP. Support for dental HIV transmission. Nature
11. Thompson JD, Higgins DG, Gibson TJ. CLUSTAL W improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucl Acids Res
12. Saitou N, Nei M. The neighbor-joining method. A new method for reconstructing phylogenetic trees. Mol Biol Evol
13. Felsenstein J. Inferring phylogenies from protein sequences by parsimony, distance, and likelihood methods. Methods Enzymol
14. Kimura M, Ohta T. Stepwise mutational model and distribution of allelic frequencies in a finite population. Proc Natl Acad Sci USA
15. Bélec L, Mohamed AS, Müller-Trutwin C, et al
. Genetically related human immunodeficiency virus type 1 in three adults of a family with no identified risk factor for intrafamiliar transmission. J Virol
16. Wolinsky SM, Wike CM, Korber BTM, et al
. Selective transmission of human immunodeficiency virus type-1 variant from mothers to infants. Science
17. Bobkov A, Cheingsong-Popov R, Garaev M, et al
. Identification of an env G subtype and heterogeneity of HIV-1 strains in the Russian Federation and Belarus. AIDS
18. Burger H, Weiser B, Flaherty K, Gulla J, Nguyen PN, Gibbs RA. Evolution of human immunodeficiency virus type 1 nucleotide sequence diversity among close contacts. Proc Natl Acad Sci USA
19. Kasper P, Kaiser R, Steinbeck-Klose A, Matz B, Schneweis KE. Elucidation of an HIV-1 transmission from mother to child in West Africa by sequence analysis. Zentralbl Bakteriol
20. Bobkov A, Garaev MM, Rzhaninova A, et al
. Molecular epidemiology of HIV-1 in the former Soviet Union: analysis of env V3 sequences and their correlation with epidemiologic data. AIDS
. 1994, 8
21. Frenkel LM, Mullins JI, Learn GH, et al
. Genetic evaluation of suspected cases of transient HIV-1 infection of infants. Science
22. Leitner T, Escanilla D, Marquina S, et al
. Biological and molecular characterization of subtype D, G, and A/D recombinant HIV-1 transmissions in Sweden. Virology
23. Holmes EC, Zhang LQ, Robertson P, et al
. The molecular epidemilogy of human immunodeficiency virus type 1 in Edinburgh. J Infect Dis
24. Leitner T, Escanilla D, Franzén C, Uhlén M, Albert J. Accurate reconstruction of a known HIV-1 transmission history by phylogenetic tree analysis. Proc Natl Acad Sci USA
25. Holmes EC, Zhang LQ, Simmonds P, Rogers AS, Brown AJL. Molecular investigation of human immunodeficiency virus (HIV) infection in a patient of an HIV-infected surgeon. J Infect Dis
26. Albert J, Wahlberg J, Leitner T, Escanilla D, Uhlén M. Analysis of a rape case by direct sequencing of the human immunodeficiency virus type 1 pol and gag genes. J Virol
27. Mellors JW, Bazmi HZ, Schinazi RF, et al
. Novel mutations in reverse transcriptase of human immunodeficiency virus type 1 reduce susceptibility to foscarnet in laboratory and clinical isolates. Antimicrob Agents Chemother
28. Li WH. Molecular phylogenetics: methods.
In: Molecular Evolution
. Sunderland, Massachusetts: Sinauer Associates,Inc.; 1997:99-147.
29. Saitou N, Imanishi T. Relative efficiencies of the Fitch-Margoliash, maximum-parsimony, maximum-likelihood, minimum-evolution, and neighbor-joining methods of phylogenetic tree construction in obtaining the correct tree. Mol Biol Evol
. 1989, 6
30. Arnold C, Balfe P, Clewley JP. Sequence distances between env genes of HIV-1 from individuals infected from the same source: implications for the investigation of possible transmission events. Virology
31. Fang G, Burger H, Grimson R, et al
. Maternal plasma human immunodeficiency virus type 1 RNA level: a determinant and projected threshold for mother-to-child transmission. Proc Natl Acad Sci USA
32. Burns DN, Landesman S, Wright DJ, et al
. Influences of other maternal variables on the relationship between maternal virus load and mother-to-infant transmission of human immunodeficiency virus type 1. J Infect Dis
33. Cao Y, Krogstad P, Korber BT, et al
. Maternal HIV-1 viral load and vertical transmission of infection: The Ariel Project for the prevention of HIV transmission from mother to infant. Nature Med
34. Dickover RE, Garratty EM, Herman SA, et al
. Identification of levels of maternal HIV-1 RNA associated with risk of perinatal transmission. Effect of maternal zidovudine treatment on viral load [see comments]. JAMA
35. Apetrei C, Descamps D, Panzaru C, Duca MC, Simon F, Brun-Vézinet F. Plasma HIV-1 load and nosocomial transmission in Romanian children [letter]. AIDS
36. Wright JJ, Birx DL, Wagner DK, Waldmann TA, Blaese RM, Fleisher TA. Normalization of antibody responsiveness in a patient with common variable hypogammaglobulinemia and HIV infection. N Engl J Med
37. Morell A, Barandun S, Locher G. HTLV-III seroconversion in a homosexual patient with common variable immunodeficiency. N Engl J Med
38. Webster ADR, Lever A, Spickett G, Beattie R, North M, Thorpe R. Recovery of antibody production after HIV infection in “common“ variable hypogammaglobulinaemia. Clin Exp Immunol
39. Noroski LM. Human immunodeficiency virus infection in a patient with primary immunodeficiency. Ann Allergy Asthma Immunol.
40. Bryant A, Calver NC, Toubi E, Webster ADR, Farrant J. Classification of patients with common variable immunodeficiency by B cell secretion of IgM and IgG in response to anti-IgM and interleukin-2. Clin Immunol Immunopathol
41. Lane HC, Masur H, Edgar LC, Whalen G, Rook AH, Fauci AS. Abnormalities of B-cell activation and immunoregulation in patients with the acquired immunodeficiency syndrome. N Engl J Med
42. Gutierrez MG, Kirkpatrick CH. Progressive immunodeficiency in a patient with IgA deficiency. Ann Allergy Asthma Immunol
43. Alter MJ, Ahtone J, Maynard JE. Hepatitis B virus transmission associated with a multiple-dose vial in a hemodialysis unit. Ann Intern Med
44. Oren I, Hershow RC, Ben-Porath E, et al
. A common-source outbreak of fulminant hepatitis B in a hospital. Ann Intern Med
45. Widell A, Christensson B, Wiebe T, et al
. Epidemiologic and molecular investigation of outbreaks of hepatitis C virus infection on a pediatric oncology service. Ann Intern Med
46. Zeuzem S, Teuber G, Lee JH, Rüster B, Roth WK. Risk factors for the transmission of hepatitis C. J Hepatol
47. Allander T, Medin C, Jacobson SH, Grillner L, Persson MAA. Hepatitis C transmission in a hemodialysis unit. J Med Virol
48. Robert LM, Chamberland ME, Cleveland JL, et al
. Investigations of patients of health care workers infected with HIV. Ann Intern Med
49. Jaffe HW, McCurdy JM, Kalish ML, et al
. Lack of HIV transmission in the practice of a dentist with AIDS. Ann Intern Med