Delays in HIV diagnosis and associated factors among patients presenting with advanced disease at a tertiary care hospital in Beijing, China

Delayed diagnosis of HIV infection is associated with advanced immunosuppression and increased risk of onward transmission. Little data exists regarding factors associated with diagnostic delays among patients presenting with advanced HIV disease in China. Medical records of patients with HIV/AIDS hospitalized at a 2000-bed tertiary hospital in Beijing, China between 1997 and 2012 were retrospectively reviewed. Demographic and clinical data of patients newly diagnosed with HIV at the hospital were abstracted. Patient characteristics, disease parameters, and the time interval between the first medical visit and the visit leading to HIV diagnosis were compared among three periods: 1997–2002, 2003–2008 and 2009–2012. Chi-square, Kruskal-Wallis and logistic regression analyses were used as appropriate. A quarter of patients (72/279) were newly diagnosed with HIV at the hospital, consisting of 11, 29 and 32 patients in 1997–2002, 2003–2008 and 2009–2012 respectively. The median time delay between the first medical visit and the visit leading to HIV diagnosis decreased over time from 91 days among patients diagnosed before 2002, to 75 days between 2003 to 2008, and 39 days after 2009 (p = 0.036). However, the median CD4+T cell count at diagnosis was 26 cells/μL [interquartile range 3–132 cells/μL] in 1997–2002, and remained unchanged across time intervals. Forty-one (57%) patients had AIDS-defining conditions and Pneumocystis jiroveci pneumonia was the most common opportunistic infection (31 cases). These results reveal persistent missed opportunities for timely HIV testing among patients with advanced disease. Strategies for promoting early HIV testing in healthcare settings are needed in China.

Introduction Delayed HIV diagnosis is associated with advanced immune compromise, increased risk of suboptimal responses to antiretroviral therapy (ART) [1], missed opportunities for preventing onward transmission and increased costs for the healthcare system [2][3][4]. There is consensus that early diagnosis of HIV infection and entry to care are essential for controlling HIV prevalence. However, a substantial proportion of HIV-infected individuals are still diagnosed at a late stage of disease worldwide. In the US, among persons aged 13 years and older, a quarter of infections were classified as having AIDS at the time of diagnosis [5]. In Europe, 54% of newly diagnosed HIV-infected individuals were late presenters as defined by having CD4 + T cell counts below 350cells/μL or an AIDS-defining illness within six months of HIV diagnosis [6]. Thus, identifying the factors associated with delayed diagnosis and attempting early testing is of great importance.
Recent studies in developed countries have demonstrated that non-traditional risk groups including women, heterosexual individuals and older adults are more likely to be tested later in the course of HIV infection [7][8][9]. These data provide important information for policymaking regarding HIV screening strategies in those settings. In China, considerable resources have been invested for HIV testing and care. Parallel systems have been implemented through the Chinese Centre for Disease Control and Prevention (China CDC) and public-sector healthcare facilities to focus on the public health and patient care-related aspects of HIV infection, respectively [10]. According to data from the Chinese national epidemiology database, from January 1, 1985 to December 31, 2009, 41% of newly identified individuals were late presenters at the time of diagnosis [11]. Notably, 30% of new HIV diagnoses were made within the hospital system as opposed to voluntary counselling and testing (VCT) sites of the China CDC system [11]. A study conducted in South China revealed that patients diagnosed in hospital settings have a significantly higher proportion of advanced HIV disease compared with those diagnosed outside of the hospital system [12]. It is likely that these individuals had presented for care at other hospitals or clinics prior to being diagnosed, however little data exist regarding the length of time between first presentation for care and the ultimate date of HIV diagnosis or the clinical characteristics that such patients with advanced disease most commonly present with.
Here, we report a study carried out at a tertiary care hospital in Beijing, China where the vast majority of patients admitted have complex or severe disease requiring multidisciplinary management or surgical intervention, providing a unique opportunity to study the temporal delay from initial presentation for care to eventual HIV diagnosis among patients with advanced disease.

Study design
This retrospective study was conducted from March 2015 to January 2016 at Peking Union Medical College Hospital (PUMCH), a 2000-bed tertiary care center in Beijing, China. The hospital provides medical services to patients from local regions (Beijing, Tianjin and surrounding Hebei province) as well as those referred from other provinces and municipalities of China. All hospitalized HIV/AIDS patients were identified based upon a list generated from the electronic database of the Medical Records Department at PUMCH. Hospital records of these patients were reviewed and eligible participants were selected according to the inclusion criteria of this study. Data from eligible patients were de-identified and entered into an electronic database which was completely anonymous. After data collection was completed, the

Data collection
All identifiable health care-related visits, including outside medical visits and PUMCH visits, were abstracted from the history of present illness of their hospital records. Visits were divided into two categories as defined previously by Kuo et al: diagnostic health care visits (DHVs), which were visits at PUMCH that directly led to the HIV diagnosis, and previous health care visits (PHVs), which included all other documented inpatient or outpatient medical visits for related symptoms prior to the DHVs [13]. The dates of the DHV and first PHV were recorded for each patient and the length of time between the first PHV and the DHV was calculated. For patients who had PHVs at PUMCH prior to their DHVs, the date of the first PUMCH PHV was also recorded. Patient demographics at the time of diagnosis including age, gender, place of residence and occupation, were abstracted. Variables collected from the history of present illness included any of the following symptoms or events: fever, unexplained weight loss, recurrent or chronic diarrhea, recurrent pneumonia, oral candidiasis, vision impairment, and varicella zoster infection. Physical examination findings collected from the DHVs included fever (! 37.3˚C), thrush, generalized lymphadenopathy, and hepatosplenomegaly. Laboratory data collected included presence of anemia (hemoglobin 120 g/L for men, 110 g/L for women), leukopenia (white blood cell count 4 000 cells/μL), lymphopenia (< 1 000 cells/μL), and thrombocytopenia (platelet count 100 000 /μL). Hepatitis B surface antigen (HBsAg) and hepatitis C virus antibody (anti-HCV), CD4 + T cell count and plasma HIV RNA level at the time of diagnosis were also abstracted from the medical records.

Statistical analysis
Descriptive analyses of patient characteristics at the time of diagnosis were performed. Continuous variables were expressed as medians (interquartile range, IQR) and compared across the three time intervals using the Kruskal-Wallis H Test. Categorical variables were expressed as frequencies and compared using the Chi-square or Fisher's exact test, as appropriate. To identify factors associated with prolonged diagnostic delay from the time of the first PHV to the DHV-defined as >63 days (the observed median value)-separate univariate logistic regression models were used to assess the odds ratio (OR) for each candidate variable, including gender, place of residence, occupation, time interval, CD4 + T cell count, number of symptoms/ events documented in the history of present illness, number of findings on physical examination, and presence of an AIDS-defining illness. Multivariable logistic regression models were then developed using variables with p-value < 0.15 in the univariate models. Backward stepwise elimination was used to fit the multivariable models. Statistical analysis was performed using SPSS 20.0 (IBM Corporation, Armonk, New York, USA) and GraphPad Prism 6.0 (GraphPad Software, Inc. La Jolla, CA, USA). A p-value < 0.05 was considered statistically significant. CD4 + T cell counts were available for 71 (99%) patients at diagnosis. The median CD4 + T cell count was only 24 (IQR 6-101) cells/μL (Table 1). Nearly two-thirds of patients (46/71, 65%) had a CD4 + T cell count below 50 cells/μL. Only 9 (13%) patients had a CD4 + T cell count above 200 cells/μL. Initial plasma HIV RNA levels were available for 49 (68%) patients with a median value of 5.08 (IQR 4.51-5.46) log10 copies/mL. Neither CD4 + T cell counts nor HIV RNA levels differed across the three time periods. More than half of patients (41/72, 56.9%) had at least one AIDS-defining condition. Pneumocystis jiroveci pneumonia (PCP) was the most common opportunistic infection (31 cases), followed by active tuberculosis (7 cases), cytomegalovirus retinitis (5 cases), esophageal candidiasis (3 cases) and pulmonary candidiasis (1 case). The distribution of AIDS-defining conditions was also similar among the three groups. Patients with AIDS-defining conditions had significantly lower median CD4 + T cell counts [12 (4-52) cells/μL] compared to those without [49  cells/μL, p = 0.003]. Among patients with AIDS-defining conditions, most documented diagnoses during their PHVs were community-acquired pneumonia (28 cases), whereas for patients without AIDSdefining conditions, 15 (48%) had an undetermined diagnosis during their PHVs.

Delay in time from the first PHV to DHV and associated factors
The overall median delay from the first PHV to DHV was 63 (IQR 26-135) days (Table 3) (Table 4). Presentation with > two symptoms or events in the history of present illness (OR 7.16, 95% CI 2.00-25.60, p = 0.002) was also significantly associated with prolonged diagnostic time delay. Prolonged time delay was not found to be significantly associated with other factors including place of residence, occupation, gender, CD4 + T cell count, number of findings on physical examination at the DHV, and presence of an AIDS-defining condition.

Discussion
This study conducted at a large tertiary care hospital in Beijing, China is the first, to our knowledge, to quantify the delay in time from initial presentation to the health care system and ultimate HIV diagnosis in patients with advanced HIV/AIDS. Our data show that although the We also found that although a large proportion of patients had clues for immune compromise, they had not been offered HIV testing during previous medical visits. These results identify missed opportunities for more timely HIV testing in medical settings, which have clinical consequences [1,14] and can negatively impact overall public health [2][3][4].
As of the end of 2014, 295 358 patients in China were receiving ART through the NFATP and 96.5% of the counties/districts across the country had qualified HIV testing labs within various medical and healthcare facilities [15]. Although the threshold CD4 + T cell count for ART initiation in the NFATP was raised from 200 cells/μL in 2002 to 350 cells/μL in 2008, over 50% of patients initiating ART between 2010 and 2011 still had a pretreatment CD4 + T cell count below 200 cells/μL [16]. Another study demonstrated that delayed ART and pretreatment CD4 + T cell count < 50 cells/μL were the strongest risk factors for overall HIVrelated mortality among adults in the NFATP [11]. Taken together, an urgent need exists for earlier HIV diagnosis, both to prevent individual morbidity and mortality from severe immunosuppression, and to better control HIV transmission in China.
One reason underlying missed opportunities for earlier diagnosis may be a lack of routine HIV screening by clinicians when patients present for medical care, even when associated symptoms are present. Since 2003, client-initiated VCT has been the primary approach to HIV screening in China. In 2007 the World Health Organization (WHO) began recommending provider-initiated HIV testing and counseling (PITC) strategies to expand access to HIVrelated services [17]. Subsequently, studies in China have demonstrated that PITC is well accepted among tuberculosis patients (99%) [18] and male attendees of sexually transmitted diseases (STD) clinics in China (91%) [19]. However, a survey of six public STD clinics in South China demonstrated that in practice, physicians providing STD care offered HIV testing to only 28% of their patients [20]. Low perceived HIV prevalence and absence of a formal recommendation in current guidelines were the most common physician-reported barriers to offering HIV testing [20]. These results highlight the under-recognition of HIV infection by clinicians in primary care settings in China, even among higher risk populations, and the need for efforts to increase awareness and encourage screening. Compared with VCT, PITC strategies rely on providers to offer testing. Therefore, more research about clinician attitudes, behaviors and practices is essential in order to identify feasible PITC strategies to reduce barriers to testing in healthcare facilities.
Furthermore, over the study period, there has been a notable shift in prevalent HIV genotypes in China. The proportion of circulating recombinant strain (CRF) 01_AE has increased significantly, especially in the population of patients acquiring HIV through sexual transmission, and CRF01_AE has now become the most prevalent genotype since 2006 [21]. Studies have shown that compared with previous predominant subtypes (for example, CRF07_BC and CRF08_BC) the CRF01_AE strain is associated with more rapid declines in CD4 + T cell counts [22,23]. This biological characteristic of the virus might explain, in part, the persistently low CD4 + T cell counts observed among our patients over time despite the decrease in length of diagnostic delay, and further underscores the importance of strategies to promote timely HIV testing in clinical settings.
Several limitations of this study warrant mention. First, due to resource constraints, our study sample was limited to patients admitted at a single tertiary care hospital in Beijing, and did not include patients from other tertiary hospitals. Therefore, the volume of patients newly diagnosed with HIV during the study period was not large. However, it is notable that 63% of the patients included in our study were from outside of the Beijing/Tianjin/Hebei region.
Second, patients diagnosed in the outpatient clinic at PUMCH were not included because detailed outpatient medical records were not available across all three time periods. As a result, our sample represents more complex and advanced cases of HIV and cannot be generalized to patients with asymptomatic or mild HIV infection. Nevertheless, over the study period we found participants were increasingly from urban centers, were more likely to acquire HIV through sexual modes of transmission, and had decreasing rates of Hepatitis C co-infection.
These key epidemiologic features are consistent with national HIV trends [15]. Therefore, despite the limitations of our study, we believe our overarching findings underscore the diagnostic challenges that remain among individuals with advanced HIV in China.

Conclusions
Temporal delays in diagnosis among patients with advanced HIV have improved significantly over a 15-year period. In this study, we identified persistent missed opportunities for timely HIV testing in healthcare settings despite increasing general awareness of HIV among patients and providers, and increased availability of HIV-related services over the past decade. Targeted strategies for promoting uptake of early HIV testing practices in healthcare settings are imperative to change this phenomenon in China, and may benefit from an increased emphasis on provider-initiated screening. Future studies should also focus on identifying pragmatic factors influencing the successful implementation and uptake of such screening strategies including feasibility, patient and provider acceptability, and provider prioritization.
Supporting information S1 File. The data set of this study. The participant-level data that underlies the results of this study.