Analyses were performed for all five MSAS outcome scales (number of symptoms, total score, global distress scale, physical scale, and psychological scale), three BPI scales (worst, ave[r]
(1)Hanoi, 2012
Integration of palliative care into routine HIV outpatient services
Findings from an intervention evaluation in Quang Ninh, Viet Nam
(2)(3)Acknowledgements
This study would not have been possible without the stalwart support and commitment from the heads of the district hospitals in Van Don and Cam Pha, the out-patient clinic (OPC) staff and the many patients who participated in this study Dr Luong Ngoc Khue, head of the Viet Nam Administration for Medical Service, was a champion of the study from the very beginning
A number of individuals from the FHI 360 team contributed to the study Dr Tran Vu Hoang was the initial co-PI followed by Dr Nguyen Cuong Quoc Drs Stephen Mills, Dr Phinh Vu Ngoc and Dr Nguyen Thi Thanh Ha were instrumental in providing inspiration and guidance throughout the course of the study Thanks goes to Drs Nguyen Cuong and Tran Thi Thanh Ha who provided research administrative support Dr Rachel Burden offered important insights into the development of the survey instrument
Dr Tuan Tran and the Research and Training Centre for Community Development (RTCCD) team are recognized for their quality work in data collection, entry and cleaning, and preliminary analysis Nguyen Nguyen Thi Trang and the Life Centre provided support with the qualitative interviews and preliminary analysis Dr Gaston Arnolda conducted analysis of the multiple rounds of data collection and contributed to the drafting of this report Thanks also goes to Daniel Levitt who edited and formatted the report
The United States Agency for International Development (USAID) and the US President’s Emergency Plan for AIDS Relief (PEPFAR) are recognized for their financial support of the study and commitment to quality HIV care
Report prepared by
FHI 360 Kimberly Green Nguyen Cuong Quoc Vu Ngoc Phinh Nguyen Thi Thanh Ha Consultant Gaston Arnolda
Study team
(4)6 Executive Summary
7 1 Introduction
8 2 Methods
8 9 11 11 13 15 Conceptual framework Site selection
Patient selection and ethical approval Standard care and intervention Data collection and entry Data cleaning
Outcome measures Analysis
16 3 Participant flow
18 4 Descriptive characteristics at baseline
18 19 20 25 26 27 28 29 Demographic characteristics Clinical history
Use of services
Symptoms and support Pain
Depression and anxiety
Social support, stigma and discrimination and service needs Drug and alcohol use and service needs
Summary
30 5 Survival analysis in the post-intervention period
30 31 32
Survival without serious illness Any loss to follow-up
Summary
32 6 Change over time
32 34 35 35 36
Physical and psychological symptom outcomes Pain outcomes
Depression and anxiety outcomes Social support outcomes
Summary
36 7 PLHIV and health care worker perceptions of the intervention
40 8 Limitations
41 9 Conclusion and recommendations
(5)Contents (cont.)
44 Annex
44 46 49 51 52 53
Annex A: Methods for multivariate model
Annex B: Cross-sectional multivariate model results Annex C: Covariates included in longitudinal models Annex D: Symptom screening tool
Annex E: Pain management poster
Annex F: Depression and anxiety screening tool
(6)ExECutivE SummAry
Palliative care aims to improve the quality of life of people living with life-threatening illnesses, and their families, from the point of diagnosis onwards In HIV policy and service provision, there is a lack of consensus regarding when to provide palliative care, and how to operationalize palliative care in HIV service delivery This report presents findings from a study that examined the degree to which integration of palliative care services into routine HIV patient care impacted clinical outcomes The study suggests that the intervention resulted in the following:
Fewer symptoms: The intervention was associated with a lower mean of symptoms and prevented the worsening of symptoms over time as measured by four outcomes in the
Memorial Symptom Assessment Scale (number of symptoms, total score, global disease index
and psychological score)
Less pain severity: There were greater reductions in worst pain and usual pain at the intervention clinic versus the control clinic, but no reductions in the way pain interferes with daily life, as measured by the Brief Pain Inventory
Reduction in depression: The intervention resulted in a reduction in depression, but not anxiety, as measured by the Hopkins Symptom Checklist - 25
Had no impact on functional social support as measured by the Functional Social Support
Questionnaire
Reduced mortality and serious illness (though without statistical significance), but did not impact overall loss to follow up
The study was conducted in two HIV outpatient clinics (OPCs) in Quang Ninh Province, in northeast Viet Nam Van Don District OPC (345 patients) as the intervention site, and Cam Pha District OPC (477 patients) as the control The OPC with the larger client population was selected as the control based on the initial analysis plan The multifaceted intervention at Van Don OPC sought to: improve provider skills in palliative care, establish systematic palliative care screening and care, increase use of essential palliative care medicines, engender support for palliative care among OPC leaders, and empower patients to communicate physical and psychosocial needs to health providers The ultimate aim of the intervention was to decrease prevalence and severity of pain and other symptoms, reduce rates of depression and anxiety, and increase social support in PLHIV
The study involved four rounds of data collection using a standardized survey instrument Given the quasi-experimental nature of the study, it included a double baseline and two post-intervention measures Baseline interviews and record reviews were undertaken in both OPCs, commencing December 2008/January 2009; the intervention in Van Don initiated in June 2009 Final interviews were conducted in June 2010, 12 months after the commencement of intervention In addition, qualitative interviews were conducted with PLHIV and health care workers to assess their perceptions of the intervention’s acceptability, accessibility and feasibility
(7)Results from the study suggest that integration of palliative care into HIV OPC in Viet Nam has a positive effect on the physical and psychological health of PLHIV It is recommended that the approach be adapted for use in other HIV clinics in Viet Nam, particularly those with a similar client profile In addition, the intervention should be tested in other contexts, ideally using a cluster-randomized comparison comparing multiple intervention and control groups
1 introduction
Palliative care is a branch of medicine that aims to improve the quality of life of people living with life-threatening illness, and their families In the HIV policy and service provision arena, there is a lack of consensus regarding when (at what point of disease progression) to provide palliative care, and how to operationalize palliative care in HIV service delivery
Viet Nam is one of the few developing nations in Asia with a national palliative care program and one that is inclusive of people living with HIV[1-3] In 2005, the Viet Nam Ministry of Health (MoH) requested support from FHI 360, Harvard University and other PEPFAR partners to assist in the assessment of the palliative care needs of PLHIV and people with cancer, and to use the findings as a basis for the development of clinical guidelines and a national palliative care program Even with the national guidelines, there was a lack of information on how to integrate palliative care into routine HIV outpatient care services
In response to the need for evidence on how best to operationalize palliative care in HIV outpatient clinics (OPCs), this study was conducted in two HIV OPCs in Quang Ninh Province, northeast Viet Nam The selected OPCs were located in Cam Pha Township and Van Don District As shown in Figure 1, Cam Pha Township is a coastal region to the northeast of the provincial capital, Ha Long City, while Van Don District is made up of a series of islands immediately to the east of Cam Pha Township Cam Pha had an estimated population of 165,842 and Van Don District had an estimated population of 39,157 in 2003
(8)These two clinics were selected since they were in adjacent districts and were similar in terms of key patient characteristics The study compared standard HIV care in Cam Pha with an intervention in Van Don that sought to formalize integration of palliative care into routine care in the OPC Changes in four palliative care outcomes were measured: pain, symptoms, depression and anxiety, and social support The study was conducted between late 2008 and the middle of 2010
2 methods
The study is a non-randomized comparison of a pair of intervention and control sites with four data collection points In the experimental arm, there were two baseline assessments prior to the intervention, and two post-intervention assessments In the control arm, there were four assessments timed to match the assessments in the experimental arm, but with no intervention Qualitative data was also collected to assess service acceptability, accessibility and feasibility Conceptual framework
The study was based on a conceptual framework that incorporated patient care needs, barriers to HIV palliative care, the tested intervention, and expected results Figure 2.1 illustrates the framework that guided the study Patient outcomes (pain, symptoms, depression and anxiety, and social support) were measured using standardized psychometric tools while mediating factors were assessed through in-depth interviews and focus group discussions with PLHIV and health care workers
Figure 2.1 HIV palliative care integration framework
Accessibility: Services are being provided and offered to PLHIV
Accessibility: PLHIV and health workers find palliative care services acceptable
Feasibility: Intervention is perveived to be “doable” by health providers
Conceptual framework: HIV palliative care integration in Viet Nam
Reduced prevalence
and severity of pain Reduced prevalence and severity of other symptoms
Reduced prevalence and severity of depression and anxiety
Mediating process factors
Clinician related Medicine assess Systems related Client related
Components for integration of palliative care in outpatient services
Palliative care needs in Viet Nam
Pain relief, treatment of symptoms and medicine side effects, care/treatment of mental health problems (e.g., depression and anxiety, social support (both inter-relational and materal) across all stages
1 HIV provider skills and belief in importance of palliative care Systematic palliative care assessnent/screening and care Increased access to/use of essential palliative care medicines Support for/commitment to palliative care by leadership PLHIV awareness/skills in communicating symptoms to OPC staff
Reduced prevalence and severity of social problems
(9)Site selection
OPCs in two adjacent administrative regions were chosen as the sites for this study The larger site was purposively selected to act as the control, so that a larger number of potential matches were available for analysis by Propensity Score Matching methods This report focuses on a description of the participants at baseline, and analysis of change over time in the two OPCs
Patient selection and ethical approval
Participation was restricted to patients 18 and over enrolled at the OPC at the time of baseline data collection, regardless of whether they had commenced ART By protocol, if a participant missed any data collection round, they were not allowed to participate in any future round in order to assess the full impact of the intervention
Basic demographic information was collected on all individuals who declined to participate in baseline data collection, and the reason for loss to follow-up was collected on each individual that ceased participation at any time
The protocol, site-specific informed consent forms, participant education and recruitment materials, and other requested documents were reviewed and approved by the institutional review boards (IRBs) of the Vietnamese Medical Association (VMA), the London School of Hygiene and Tropical Medicine, and FHI 360
Standard care and intervention
Prior to the intervention, standard care in the two clinics included palliative care as an informal element of care However, it did not have a specific focus or standard operating procedures with specific goals to ensure routine screening and implementation of palliative care A multifaceted intervention at Van Don OPC, which commenced after second baseline data were collected, sought to enhance palliative care by:
1 Improving HIV provider skills in palliative care and enhancing their belief in its importance Establishing systematic palliative care assessment/screening and care
3 Increasing access to and use of essential palliative care medication
4 Engendering support for and commitment to palliative care by hospital and clinic leadership Empowering the client population to communicate physical and psychological concerns to
health care workers
(10)Table 2.1 Standard versus enhanced care
Care site Cam Pha HIV OPC
Standard care services Intervention servicesVan Don HIV OPC
Outpatient
clinic • OI/TB treatment• ART & ART adherence support
• Ad hoc palliative care • HTC/PMTCT
• Nutrition support • Some social support
(supplemental food, education support and links to social welfare schemes - mostly for PLHIV with children)
Standard care services +
• Routine assessment and treatment of pain and other symptoms
• Treatment for moderate-to-severe psychiatric morbidity as indicated • Referral to mental health counselors • Referral to CHBC team and/or PLHIV support group for further support
CHBC team • Assessment and treatment of
mild pain and other symptoms using home-care kit
• Self-care skills building
• ART adherence and care of side-effects
• Basic emotional support • Social welfare – income
generation, enrollment in government programs
• Ad hoc links for spiritual support • End-of-life care
• OVC services
Standard care services +
• Linked follow-up care to PLHIV identified in the OPC with moderate to severe pain, symptoms and psychosocial problems
• Use of simple pain, symptoms and psychosocial problems scales to determine emergencies or lack of improvement in clients, and referral to OPC
• Training of PLHIV and families regarding palliative care needs and services
PLHIV
sup-port group • Monthly meetings• Self-care education at meetings
• Income generation activities • Ad hoc home-visits
Standard care services +
• Education of members regarding palliative care problems and services offered at the clinic
• Linked referrals for PLHIV identified with psychosocial problems who agree to join group
Lay mental health counseling
Did not exist + Screening for depression and anxiety + Assessment of levels of support from family and friends, support concerns and material needs
+ Provision of psycho-education and supportive counseling
+ Treatment of depression and anxiety + Referrals for clients with severe
depression and anxiety
(11)Data collection and entry
Data were collected through interviews and by review of medical records FHI 360 worked with a local research organization based in Hanoi, the Research and Training Center for Community Development (RTCCD), to select and train a team of interviewers Each interviewer was trained on research ethics, confidentiality, HIV stigma and discrimination, documentation of illnesses and death, and in how to administer the questionnaire Data entry clerks were trained to enter study-specific data in the database Ten percent of all interviews were redone to check for quality In addition, all surveys were double entered in a Microsoft Access database; any inconsistencies were discussed with the Primary Investigator (PI) and co-PI
Baseline interviews and record reviews were undertaken in both OPCs late December 2008 and January 2009, and were repeated 3-6 months later The intervention commenced in Van Don in June 2009 Two rounds of follow-up data collection were conducted: the first in December 2009 and the second in June 2010, 12 months after the commencement of intervention
The Life Center assisted FHI 360 to conduct qualitative interviews with PLHIV and health care workers from Van Don and Cam Pha district hospitals Two rounds of data collection were conducted, once at baseline (January 2009) and once after the intervention had been implemented in Van Don (December 2009/January 2010) PLHIV support group leaders were interviewed in focus group discussions to assess their perception of PLHIV palliative care needs and to gauge their opinions of the intervention In-depth interviews were held with PLHIV who experienced typical palliative care problems (e.g pain, anxiety) to learn more about their experiences and expectations of care from the OPC, and perceptions of the intervention A total of 41 PLHIV were interviewed at baseline, and 35 after the intervention had been implemented See table 2.2 for details
Table 2.2 Qualitative sample for PLHIV
Interview Type/Time Pre-test (Time 1) Post-test (Time 3)
Female Male Female Male
Focus group discussions Cam Pha (2 groups) 7 7
Van Don (2 groups)
Semi-structured in-depth
interviews Cam Pha 0
Van Don 10 10
Total included in analysis 20 21 18 17
In addition, a total of 23 health care workers were interviewed from the Cam Pha and Van Don district hospitals and HIV OPCs at baseline, and 21 in December 2009/January 2010 Like PLHIV, health care workers were asked to provide information on their perception of palliative care needs among their patients Post-intervention, these health care workers were asked to assess the acceptability, accessibility and feasibility of the intervention
Data cleaning
(12)Information relating to dates
Dates were checked to ensure that they were consistent between rounds of data collection (e.g., date of first commencement of ART), and corrections were made as necessary Dates that should have changed from one round to the next (e.g., date of interview, date of most recent CD4 count) were checked for accuracy and corrected, or identified as missing
Commencement of ART and assessment of adherence
ART status and use was assessed through patient self-reports, and through examination of medical records Examination of consecutive rounds of data allowed correction of errors in recording whether or not a patient had initiated ART, and the date of commencement The team identified twenty-two records of clients who, according to medical records, had started ART, but who did not self-identify as ART clients These participants were classified as on ART for this report
Injection drug use
At each interview session, participants were asked if they had ever injected drugs and, if so, whether they had injected drugs in the previous months This led to some contradictory responses when data from different rounds were compared These were resolved using the following rules:
Forward correction: An individual who reported that he/she had previously injected drugs in one round was classified as an ‘injection drug user’ (IDU) in that round and all subsequent rounds (i.e., the first self-identification as an IDU was presumed correct)
Backward correction: An individual who stated that he/she was an IDU for the first time in interview rounds 2-4 was classified as an IDU in the round immediately previous if: he/she stated that he/she had not injected drugs in the previous months AND there were fewer than 183 days between the two rounds (i.e., it was assumed that he/she was honest about recent injection drug use, but had been dishonest about drug use in the previous round) The corrections resulted in the re-classification of 88 records - 78 records (49 individuals) were reclassified using the forward correction rule, and 10 records (10 individuals) were reclassified using the backward correction rule Of these, four individuals were re-classified by both forward and backward correction rules, hence data for a total of 55 individuals were adjusted
CD4 and WHO clinical stage
At each round, individual medical records were examined for specific data For CD4 and WHO clinical stage measures, data collectors used the most recent value in the medical record, and the date of the value This inevitably resulted in repetition of some values in cases where no new reading had been conducted between interview rounds It also occurred where, on occasion, an earlier reading of the value was recorded in a subsequent round To account for these, all individual records were examined, repetitions were set to missing, and/or other errors corrected
Physical functioning scale adapted from the MOS SF-36
(13)Outcome measures
Four separate types of outcome scales were used in this study, each with one or more sub-scales Physical and psychological symptoms (the Memorial Symptom Assessment Scale (MSAS)) The MSAS is a widely validated, used and recommended palliative care outcome measure for symptoms in people with cancer and HIV[4-7] Because it had not been previously validated in Vietnam, the language and terms were first reviewed by health care workers, PLHIV and local researchers and deemed as clear, acceptable and relevant The MSAS was translated into Vietnamese, and reviewed and revised by health care workers and researchers It was then pilot tested with PLHIV, revised and translated back into English
The scale is made up of 32 standard questions, but the analysis team deemed the questions related to weight and bloating poorly translated and removed them from the analysis As with other HIV studies that use this scale[8], six prevalent HIV symptoms, according to Vietnamese clinicians and PLHIV, were added to the original symptom list They included the following: headache, backache, problems with vision, tooth pain/problems, painful skin rash, and genital itchiness/pain For each symptom, respondents indicated if the symptom presented during the past month and, if so, they were asked the following:
The frequency of that symptom (scored by Likert scale: = no symptom, = rarely, = occasionally, = frequently, and = constantly) [only asked in 23 questions]
The severity of that symptom (scored as for frequency)
How much the symptom distressed or bothered them (scored by Likert scale: = no symptom, 0.8 = causes no distress, 1.6 = a little bit, 2.4 = somewhat, 3.2 = quite a bit, and 4.0 = a lot) The MSAS produces three validated scales (numbers 1-3 below) Three other overall measures (4-6 below) found in the literature are reported in this study:
1 The MSAS Global Distress Index (MSAS-GDI) calculated as the mean of 10 items: the frequency scores for four psychological items and the distress reported on six physical items The MSAS Psychological Symptom Subscale (MSAS-PSYCH): the mean of frequency, severity
and distress for each of six psychological symptoms (18 items in total)
3 The MSAS Physical Symptom Subscale (MSAS-PHYS): the mean rating of frequency, severity and distress for each of physical symptoms and severity and distress of items (28 items in total) This subscale usually includes weight and bloating, but these were unavailable due to errors in translation
4 The prevalence of each symptom[8, 9]
5 The symptom burden: the average number of symptoms per patient[8, 9]
6 The MSAS-Total score: the mean rating of frequency, severity and distress for each of 23 symptoms (69 items), and severity and distress for seven symptoms (14 items, for a total of 83 items)[9]
Pain prevalence, severity and interference (the Brief Pain Inventory (BPI))
(14)The inventory begins by asking the respondent to identify all locations where the respondent has had pain and then proceeds to explore two dimensions of pain:
1 The sensory dimension has four sub-elements of which two (worst, average) were used in this study Respondents score each sub-element from zero (no pain) to 10 (worst pain imaginable) Scores are classified as none (0), mild (1-3), moderate (4-6) or severe (7-10) The reaction to pain dimension is measured as interference with seven daily activities –
general interference, walking, work, mood, enjoyment, relations with others, and sleep The respondent scores each of the seven activities from zero (no interference) to 10 (complete interference) If there are measures (other than 0) in four or more of the seven activities, the mean of the items with measures indicates the degree of interference in the patient’s life[10] To align the BPI with the MSAS (discussed above), the reaction to pain dimension was measured by asking about pain in the past month, rather than in the past 24 hours Given ‘pain’ in the Vietnamese language is generally associated with the locus of pain (e.g., headache) rather than as pain alone, patients were only asked to complete the BPI questionnaire if they answered yes to one of the options provided in the following MSAS question - During the past month, did you have any of the following symptoms: pain, numbness, shooting pain, tingling in hands or feet, difficulty swallowing or pain when swallowing, headache, backache, painful skin rash, or trouble with teeth or painful teeth If a respondent answered ‘no’ to all six of these questions (i.e., no pain), he/she was classified as having a pain score of for both the sensory dimensions (i.e., no pain) and for the reaction to pain dimension (i.e., no reaction to pain since there was no pain) Depression and anxiety (Hopkins Symptom Checklist (HSCL-25))
The HSCL-25 was originally developed in the 1950s and has since been adapted and validated for use in countries across Asia, Africa, Latin America and Europe[14] It is widely used in palliative care and HIV studies and in developing country contexts[16] The scale was validated in Vietnamese populations in 1987[15] The test-retest coefficient in Vietnamese populations was 89%; sensitivity for major psychiatric morbidity was 93%, and specificity 76%
The HSCL-25 is made up of 25 questions - 10 relating to anxiety and 15 relating to depression Each of the questions is assessed on a four-point Likert scale where signifies ‘no problem’, and signifies ‘an extreme problem’ The average of all responses is calculated separately for the 10 questions on anxiety, the 15 questions on depression, and the 25 questions overall In this study, a mean score of >1.75 was classified as symptomatic for either anxiety or depression or poor overall mental health[17]
Social support (Functional Social Support Questionnaire (FSSQ))
No social support scales validated for use in Vietnamese populations were identified The research team chose the Duke University & University of North Carolina FSSQ scale as the most appropriate social support scale for this study given its use in HIV studies and in developing country contexts[16, 18] The scale was modified for the developing country context to include 10 items[16] A cultural review of the scale by Vietnamese health care workers and researchers further reduced it to seven statements - five statements related to emotional and/or affective support, and two related to material and/or instrumental support
(15)Analysis
Comparison at baseline
Comparison of baseline data included tabulations of sample sizes and proportions for categorical, median and first and third quartiles for non-normally distributed continuous data, or means and standard deviations for normally distributed continuous data For each case, the tables present overall measures (including both clinics), and measures for each clinic separately
Proportions in each clinic were compared by chi-square tests after exclusion of missing data and incorrect responses A number of outcomes in the study had skewed data, with strong floor effects; univariate comparisons were therefore performed using Wilcoxon test for comparing medians Cross-sectional, multivariate comparisons of outcomes were performed to assess whether there were differences between the clinics at baseline after adjustment for covariates Analyses were performed for all five MSAS outcome scales (number of symptoms, total score, global distress scale, physical scale, and psychological scale), three BPI scales (worst, average, and pain interference), three HSCL scales (anxiety, depression, poor mental health), and the FSSQ All were modeled as continuous variables using multiple linear regression except for the three HSCL scales, which were modeled with logistic regression, after being dichotomized (>1.75) at the threshold for defining anxiety, depression, or poor mental health Details of model-building and checking strategies are summarized in Annex A
Survival analysis
The team conducted a survival analysis to explore losses to follow-up due to death, serious illness, or any reason except transfer to another OPC The analysis included the period from the second baseline through the second follow-up round A significant difference in survival between the two clinics can be interpreted as a possible impact of the intervention Differences between Van Don and Cam Pha were explored by stratified Kaplan-Meier plots, and by proportional hazards regression to allow adjustment for risk factors The modeling procedure is specified in Annex A Change over time
(16)3 Participant flow
The flow of participants through this study is shown in Figure Virtually all registrants agreed to participate – only 8 registrants at Cam Pha (1.6%) and 2 at Van Don (0.6%) declined the invitation to participate
Figure 2: Participant flow and losses
refused (1.6%)* refused (0.6%)*
Van Don
322 repeated baseline
(78 to 181 days after Baseline 1)
291 Round
(237 to 337 days after Baseline 2)
267 Final Round
(158 to 212 days after Round 3)
439 repeated baseline
(82 to 184 days after Baseline 1)
381 Round
(242 to 342 days after Baseline 2)
349 Final Round
(153 to 206 days after Round 3)
Cam Pha 345 baseline
interviews 477 baseline interviews
transferred out 15 arrested lost contact discontinued too ill died 38 Total (8.0%)*
transferred out 20 arrested 14 lost contact discontinued too ill 14 died
58 Total (13.2%)*
transferred out arrested 16 lost contact discontinued too ill died 32 Total (8.4%)* transferred out
arrested lost contact discontinued too ill died 23 Total (6.7%)*
transferred out 16 arrested lost contact discontinued too ill died 31 Total (9.0%)*
transferred out arrested 11 lost contact discontinued too ill died 24 Total (8.2%)*
* The number of participants in the previous interview round (in each clinic) is the denominator for the total percent lost
(17)Table 1: Comparison of the 822 participants with the 10 who decided not to participate
Participants Refusals
Number (% of total) 822 (98.8%) 10 (1.2%)
% from Cam Pha 58.0% 80.0%
Median age (years) 31.0 33.5
% male 68.4% 50.0%
% single 29.8% 30.0%
% on ART 26.2% 20.0%
% clinical stage or 66.6% 60.0%
Median CD4 (mg/dL; before December 2008) 339 298
The first baseline interviews were conducted at the end of December 2008 The second baseline interview was a mean of 138 days later (~4.5 months) The intervention at Van Don was implemented starting in June 2009 and the Round interviews took place a mean of 280 days (~9.3 months) after the second baseline, with Final Round interviews conducted a mean of 182 days (~6.0 months) after Round There was no difference in the average duration between interviews for Van Don and Cam Pha clients
Of the 822 registrants who agreed to participate, 206 (25.0%) were lost to follow-up by the last round of interviews Sixty-one were lost after Round (7.4% of participants at baseline); an additional 89 (10.8% of participants at baseline) were lost between Rounds and 3; and a further 56 (6.8%) were lost between Rounds and The 206 who did not complete the study were lost for a variety of reasons Seventy-two (35%) were arrested; 60 (29%) could not be contacted; 24 transferred to other OPCs; (3%) withdrew from the study; 35 (17%) died; and (4%) were too ill to participate Figure displays the losses between interview rounds separately for each OPC, and the reasons for loss Once a participant missed an interview round, he/she was no longer part of the study, even if he/she returned to the OPC However, very few returned The percentage of participants lost during the study was consistently higher at Cam Pha than at Van Don (each percentage in Figure represents the percentage of clients who attended the preceding interview but did not return, by round) The differences in loss between the two baseline rounds are minor (6.7% lost at Van Don vs 8.0% at Cam Pha) They are more marked between the Second Baseline Round and Round (9.0% vs 13.2% respectively), and again fairly minor between Round and the Final Round (8.2% vs 8.4% respectively)
(18)4 Descriptive characteristics at baseline
This section describes the overall baseline characteristics of the 822 patients who consented to participate in the study, and examines differences between those at Van Don and Cam Pha OPCs Demographic characteristics
The majority of patients enrolled in both clinics were male (71% in Van Don vs 68% in Cam Pha, p=0.22) Clients ranged in age from 18-48; the median age in both clinics was 31 years (not statistically significant [ns]) While the age structure was similar in the two OPC client groups, Van Don clients had lower levels of education (22% with less than Grade education vs 7% in Cam Pha, p<0.0001 overall) Van Don clients were more likely to be (currently) married (61% vs 50%, p=0.006 overall), and were more likely to have children (71% vs 58%, p<0.0001 overall) See Table 2A (categorical data) and Table 2B (continuous data) for a comparison of demographic characteristics from the two OPCs
Median monthly household income for clients was about 3.5 million VND (equivalent to US$167) overall (interquartile range [IQR] of 2.0 – 5.5m VND); this did not differ between Van Don and Cam Pha clients (p=0.64) The survey also contained four questions about the adequacy of household income for various purposes (food, transport, medical care, and schooling of children) An average score across these four questions was used to estimate economic strain for each client, measured on a scale of 1-5 (1 indicates no economic strain; indicates extreme economic strain) The median score overall was 3.0, which is equivalent to lacking sufficient funds for activities ‘Fairly often’, and the IQR ranged from 2.0 (‘Once in a while’) to 4.0 (‘Very often’) There was no difference between the median scale scores of Van Don (2.8) and Cam Pha (3.0) (p=0.19)
Of all respondents, 25% reported that they had been in prison or an 05/06 center - 17% had been to prison only, 6.5% to an 05/06 center only, and 1.9% to both Van Don clients were less likely than Cam Pha clients to report detention in each of these categories, but the overall difference was not statistically significant (p=0.09) Van Don clients were less likely to have ever been detained (21%) than Cam Pha clients (28%; p=0.01)
Table 2A: Selected demographic characteristics of Van Don and Cam Pha clients *
All OPC
Van Don Cam Pha
p-value* N Median (IQR) Median (IQR) Median (IQR)
Age (years) 822 31.0(28.0-35.0) 31.0(29.0-34.0) 31.0(28.0-35.0) 0.82
Economic strain scale 822 3.0(2.0-4.0) 2.8(2.0-4.0) 3.0(2.0-4.0) 0.19
Average monthly total household
income 752
3.5m VND (US$167) (2.0-5.5m)
3.5m VND (US$167) (2.0-5.5m)
3.6m VND (US$172)
(19)Table 2B: Selected demographic characteristics of Van Don and Cam Pha clients*
All OPC
Van Don Cam Pha
p-value*
N % % %
Male 562 68.4 70.7 66.7 0.22
Education
None, or
Grades 1-5 108 13.1 21.7 6.9
<0.0001
Grades 6-9 344 41.8 44.6 39.8
Grades 10-12 255 31.0 22.6 37.1
Tertiary 115 14.0 11.0 16.1
Marital status
Married 446 54.3 60.6 49.7
0.006
Single 245 29.8 24.6 33.5
Divorced/ separated/
widowed 131 15.9 14.8 16.8
Intercourse in last month?
Yes 395 48.1 51.9 45.4
0.07
No 426 51.9 48.1 54.6
Number of children
0 300 36.5 29.3 41.7
<0.0001
1 374 45.5 46.7 44.7
2 136 16.5 22.0 12.6
3 12 1.5 2.0 1.0
Current employment status
Unemployed 250 30.5 28.7 31.8
0.63
Full time 440 53.7 54.8 52.8
Part time 130 15.9 16.5 15.4
Prison or 05/06 Center? Yes 206 25.1 20.6 28.3 0.01
615 74.9 79.4 71.7
Total 822 100.0 100.0 100.0
-* Italicized p-value indicates overall statistically significant difference between Van Don and Cam Pha Percentages calculated after removal of non-respondents
Clinical history
The team used medical records and patient self-reports to determine if a participant had commenced ART Participants were classified as having begun ART If either one of these sources indicated so Overall, 628 clients (76%) had commenced During the duration of the study, a higher proportion of Van Don clients (83%) initiated ART than those at Cam Pha (72%; p=0.0002)
(20)Among the 624 patients who had commenced ART, the median time from commencement of ART to the date of baseline interview was 17 months The median time from commencement of ART was longer at Van Don (18 months) than Cam Pha (16 months; p=0.008)
All clients were asked about their perception of their own health over the last month, reported on a five-point Likert scale from Very Good (1) to Very Poor (5) The proportion of those who rated their health as Good or Very Good was 17.7% at Van Don and 19.5% at Cam Pha (p=0.53)
Clients were asked if their health limited six activities drawn from the physical functioning scale of the SF-36 Overall, 80% reported one or more limitations There was no difference between the proportions of clients who reported these limitations at Van Don and Cam Pha, for any individual category, and overall The median score on these six activities comprises the scale score, which runs from 0-100, with indicating large limitations and 100 indicating no limitation The median scale score was 83.3 overall, and was identical for Van Don and Cam Pha clients (p=0.60)
All clients were also asked if their health impacted their ability to find work Of the 814 participants who responded, half (50%) reported a negative impact There was little difference between Van Don clients (53%) and Cam Pha clients (48%; p=0.16)
Table 3: Selected aspects of the clinical histories of Van Don and Cam Pha clients*
All OPC
Van Don Cam Pha
p-value* N Median (IQR) Median (IQR) Median (IQR)
CD4 (mg/dL) 806 337(180-524) 353(197-551) 319(165-502) 0.03
Months from testing HIV positive
to baseline interview 804 32.5(17.0-65.0) 31.0(17.0-56.0) 38.0(17.0-67.0) 0.22 Months from starting ART to
baseline interview* 619 17.0(12.0-24.0) 18.0(13.0-24.0) 16.0(9.0-23.0) 0.008 Physical functioning score from
SF-36 (adapted) 822 83.3(66.7-91.7) 83.3(66.7-91.7) 83.3(66.7-91.7) 0.60
* Italicized p-value indicates overall statistically significant difference between Van Don and Cam Pha IQR: Interquartile range
Use of services
All 822 patients were asked about their current use of services (in addition to clinical care and ART), highlighted in Table The most commonly used services were community/home-based care (CHBC) services (50% of clients), emotional counseling services for self (26%) and family (11%), and PLHIV support groups (14%) Fewer than 5% reported the use of any other service Van Don clients were more likely to report use of CHBC services (57% vs 46%; p=0.002), use of PLHIV support groups (21% vs 9%; p<0.0001), nutritional support (5.0% vs 1.9%; p=0.02), and support with transport (5.6% vs 0.9%; p<0.0001)
(21)Table 4: Use of services by Van Don and Cam Pha clients*
Type of service All
OPC
Van Don Cam Pha
p-value*
N % % %
CHBC 405 50.4 56.9 45.6 0.002
PLHIV support group 114 14.2 20.8 9.3 <0.0001
Emotional counseling for self 212 26.4 25.8 26.8 0.81
Emotional counseling for family 85 10.6 12.9 8.9 0.08
Total 804 100% 100% 100%
-* Italicized p-value indicates overall statistically significant difference between Van Don and Cam Pha; 18 missing
Symptoms and support
Thirty-six symptoms were assessed in the MSAS questionnaire used in this study The vast majority of participants (98.7% in Cam Pha and 98% in Van Don; P=0.38) reported having at least one symptom in the past month Table 5A shows the number and proportion of clients reporting each symptom overall, the proportion in each OPC, and the proportion reporting each symptom separately by ART status Participants reported the presence of many common symptoms; the most common were difficulty concentrating (52%), lack of energy (54%), dry mouth (51%), sadness (51%) and irritability (52%)
(22)Table 5A: Proportion of respondents at baseline self-reporting on 36 symptoms, overall, and separately by clinic and ART status
MSAS symptoms All
Clinic On ART?
Van
Don Cam Pha p-value*
Yes No
p-value*
N % % % % %
Difficulty concentrating 425 51.7 53.0 50.7 0.53 50.6 55.2 0.29
Pain 300 36.5 32.2 39.6 0.03 34.2 43.8 0.02
Lack of energy 443 53.9 53.0 54.5 0.72 52.1 59.8 0.07
Cough 357 43.4 44.1 43.0 0.78 41.9 48.5 0.12
Nervousness 269 32.7 34.5 31.4 0.37 32.2 34.5 0.54
Dry mouth 418 50.9 55.4 47.6 0.03 49.5 55.2 0.19
Nausea 204 24.8 23.5 25.8 0.46 24.5 25.8 0.78
Vomiting 100 12.2 11.3 12.8 0.59 12.1 12.4 0.90
Drowsiness 165 20.1 18.0 21.6 0.22 18.8 24.2 0.10
Numbness/shooting pain/tingling in
hands/feet 313 38.1 39.1 37.3 0.61 36.9 41.8 0.24
Difficulty sleeping 389 47.3 48.4 46.5 0.62 47.3 47.4 1.0
Problems urinating 83 10.1 10.7 9.6 0.64 8.6 14.9 0.01
Shortness of breath 169 20.6 18.3 22.2 0.19 19.7 23.2 0.31
Diarrhea 120 14.6 14.8 14.5 0.92 12.4 21.6 0.002
Sadness 422 51.3 51.0 51.6 0.89 50.5 54.1 0.41
Sweating 220 26.8 27.2 26.4 0.81 26.3 28.4 0.58
Worrying 406 49.4 48.7 49.9 0.78 47.9 54.1 0.14
Problems with sexual interest or
activity 162 19.7 16.5 22.0 0.051 19.6 20.1 0.92
Itching 191 23.2 17.4 27.5 0.0008 23.6 22.2 0.77
Lack of appetite 346 42.1 36.8 45.9 0.01 40.3 47.9 0.07
Dizziness 338 41.1 38.6 43.0 0.22 37.3 53.6 <0.0001
Difficulty swallowing, painful
swallowing 92 11.2 10.7 11.5 0.74 10.7 12.9 0.43
Irritability 423 51.5 50.4 52.2 0.62 51.1 52.6 0.74
Headache 393 47.8 47.0 48.4 0.72 46.7 51.5 0.25
Backache 313 38.1 39.7 36.9 0.42 35.7 45.9 0.01
Problems with vision 169 20.6 24.1 18.0 0.04 21.8 16.5 0.13
Painful skin rash 65 7.9 6.1 9.2 0.12 7.8 8.2 0.88
Trouble with teeth; painful teeth 168 20.4 19.7 21.0 0.73 18.2 27.8 0.004
Vaginal or penile discharge 313 38.1 39.1 37.3 0.46 36.9 41.8 0.24
Mouth sores 75 9.1 9.9 8.6 0.54 7.5 14.4 0.006
Change in the way food tastes 119 14.5 12.5 15.9 0.19 14.5 14.4 1.0
Hair loss 253 30.8 31.6 30.2 0.70 31.5 28.4 0.42
Constipation 213 25.9 26.1 25.8 0.94 25.8 26.3 0.93
Swelling or arms of legs 36 4.4 3.5 5.0 0.31 4.5 4.1 1.0
Changes in shape of body 49 6.0 7.0 5.2 0.37 6.7 3.6 0.12
Changes in skin 218 26.5 27.0 26.2 0.81 26.4 26.8 0.93
(23)Table 5B: Proportion of respondents at baseline self-reporting on 30 selected symptoms, and the same symptoms as recorded in the medical notes at the last clinical visit
MSAS symptoms Self-reported during baseline interview
Recorded in last medical notes at last
clinical visit*
N % N %
Pain 300 36.5 10 1.2
Lack of energy 443 53.9 0.2
Cough 357 43.4 23 2.8
Dry mouth 418 50.9 0.0
Nausea 204 24.8 0.1
Vomiting 100 12.2 0.1
Feeling drowsy 165 20.1 0.0
Difficulty sleeping 389 47.3 0.1
Problems urinating 83 10.1 0.1
Shortness of breath 169 20.6 0.1
Diarrhea 120 14.6 40 4.9
Feeling sad 422 51.3 0.0
Worrying 406 49.4 0.0
Problems with sexual interest or activity 162 19.7 0.0
Itching 191 23.2 20 2.4
Lack of appetite 346 42.1 0.4
Dizziness 338 41.1 0.1
Difficulty swallowing, painful swallowing 92 11.2 0.0
Headache 393 47.8 20 2.4
Backache 313 38.1 0.1
Problems with vision 169 20.6 0.4
Painful skin rash 65 7.9 0.1
Trouble with teeth; painful teeth 168 20.4 0.2
Vaginal or penile discharge 313 38.1 0.1
Mouth sores 75 9.1 0.2
Change in the way food tastes 119 14.5 0.0
Hair loss 253 30.8 0.0
Constipation 213 25.9 0.0
Swelling or arms of legs 36 4.4 0.1
Changes in shape of body 49 6.0 0.0
Total 822 100% 820 100%
* Two records lacked any symptoms in the medical record of the most recent clinic visit
(24)Table 5C: Median symptom (MSAS) scores, and the number of symptoms reported overall, by OPC and ART status
All Clinic On ART?
Van Don Cam Pha p-value*
Yes No
p-value* N Median (IQR) Median(IQR) Median(IQR) Median(IQR) Median(IQR)
Symptom burden 822 10
(6-16) (5-15)10 (6-15)10 0.70 (5-15)9 (6-16)11 0.01 Global Distress
Index
(10 symptoms) 822
0.8
(0.3-1.4) (0.3-1.4)0.8 (0.4-1.4) 0.280.8 (0.3-1.4)0.8 (0.4-1.6)0.9 0.02 Psychological
symptom subscale
(6 symptoms) 822
0.9
(0.3-1.5) (0.3-1.6)0.9 (0.3-1.5) 0.660.9 (0.3-1.5)0.9 (0.4-1.7)1.1 0.07 Physical symptom
subscale
(10 symptoms) 822
0.6
(0.2-1.0) (0.2-0.9)0.5 (0.2-1.1) 0.140.6 (0.2-1.0)0.5 (0.3-1.3) 0.0040.7 Total score (30
symptoms) 822 (0.3-0.9)0.5 (0.3-0.9)0.5 (0.3-1.0) 0.300.5 (0.3-0.9)0.5 (0.3-1.1) 0.010.6
* Italicized p-value indicates overall statistically significant difference between Van Don and Cam Pha IQR: Interquartile range
As shown in Table 5C, the overall median number of symptoms reported was 10; this did not differ between Van Don and Cam Pha Patients on ART reported a slightly lower median number of symptoms than those not on ART (9 vs 11; p=0.01)
The sub-scales presented in Table 5C indicate that overall distress and psychological symptoms are clinically important, while physical symptoms were less severe in this cohort A similar pattern was seen for each of the four MSAS scales The scales ranged in value from 0.5 to 0.9 overall, with no difference between Van Don and Cam Pha patients Patients on ART, however, consistently had slightly lower scale values than those not on ART; this difference was statistically significant for all scales except the MSAS psychological symptom subscale
Among the 810 PLHIV who reported symptoms, 533 (66%) sought treatment, with no difference in the proportion at each OPC (68% vs 65%, p=0.37) Self-treatment was most common, practiced by 389 patients (74% of the 533 who sought treatment), followed by treatment at the OPC (57%), treatment by family members (38%), and treatment by the CHBC team (26%) The proportion of clients who sought treatment from each of these sources did not differ between Van Don and Cam Pha
After receiving support, 26% of clients reported that they were completely recovered, 34% mostly recovered, 29% somewhat improved, and only 11% reported that their symptoms were still present or worse There was no difference in the distribution of these responses between Van Don and Cam Pha clients (p=0.34)
(25)Pain
Of the 822 clients who completed the baseline interview, 597 (73%) indicated that they had experienced pain in the previous month in response to one of seven MSAS questions, with no difference in the proportion reporting pain in Van Don (72%) and Cam Pha (73%; p=0.69)
There was no difference in the median number of pain locations identified by Van Don and Cam Pha clients (p=0.87) The proportion of clients reporting pain in different locations was: 69% in the head, 60% in the torso, 30% in the legs, 27% in the abdomen, 24% in the arms, 20% in the neck, 19% in the hands, 17% in the feet, 4% in the genitals, and 13% in other parts of the body Van Don and Cam Pha clients had similar proportions for each of the parts listed above (all p>0.05)
Clients who experienced pain were asked to identify worst and average pain on a scale from (no pain) to 10 (worst imaginable) These were then classified as no pain, mild pain, moderate pain or severe pain as shown in Table For both of these sensory dimensions of pain (worst and average) Van Don clients were more likely to report pain in the severe categories (p=0.005 for worst pain and p=0.007 for average pain; overall comparison, Fisher’s Exact Test)
Table 6: Worst and average pain over last month, overall and by OPC*
Pain rating All
OPC
Van Don Cam Pha
p-value*
N % % %
Worst pain
None 0.7 1.2 0.3 0.002
Mild 115 19.4 13.8 23.4 <0.0001 Moderate 274 46.2 46.2 46.2 0.81
Severe 200 33.7 38.9 30.1 0.08
Average pain
None 10 1.7 2.8 0.9
-Mild 502 84.9 79.3 89.0
Moderate 76 12.9 17.1 9.9
Severe 0.5 0.8 0.3
Total 593 100% 100% 100%
* Italicized p-value indicates overall statistically significant difference between Van Don and Cam Pha
Participants were also asked to describe how pain interfered with seven daily activities In all, 594 of 597 clients reported pain interference on 5-7 of the daily activities The BPI pain interference score ranged from (no interference) to 10 (maximum interference); the median was 3.2 overall (IQR: 1.1-5.3), with median interference scores similar in the two OPCs (3.3 in Van Don vs 3.1 in Cam Pha, p=0.62)
(26)Clients were asked questions about use of either morphine or codeine Thirty-nine percent reported that they were uncomfortable using these drugs for pain relief Van Don clients (44%) appeared more likely to report these concerns than Cam Pha clients (37%), but the difference was not statistically significant (p=0.09) Patients who reported these concerns gave one to five reasons for being concerned: 77% fear of addiction, 36% fear of side-effects, 21% fear of relapse, 8% fear their families would not understand, 6% fear of arrest, and 12% other fear(s)
Depression and anxiety
The proportion of clients with anxiety, depression and poor mental health was assessed at the standard HSCL threshold (>1.75)[17]:
Clinical anxiety was estimated in 25% of clients overall - in 26% of Van Don clients and 24% of Cam Pha clients (p=0.62)
Clinical depression was reported by 42% of clients overall - in 40% of Van Don clients and 43% of Cam Pha clients (p=0.39)
Poor mental health (score >1.75 across all 25 questions) was reported by 37% of clients overall - in 36% of Van Don clients and 38% of Cam Pha clients (p=0.71)
Univariate analysis of depression and anxiety by sex indicates a large disparity between female and male study participants Graph 5.4a shows that the prevalence of depression in women was 53.7% versus 36.7% in men The gap is slightly less pronounced in anxiety, where 30.3% women and 22.3% men were symptomatic Of note is the fact that depression and anxiety levels are much higher in this study than general population estimates in rural Vietnamese populations A recent study by Giang et al, presented in Graph 5.4a, found mental health distress in rural northern Viet Nam to be 5.4% (6.8% in women and 3.9% in men) (Giang et al 2010)
(27)All clients were asked if they had received emotional counseling at the OPC, during the preceding three months, when they were feeling sad Of the 813 that responded, 19% indicated that they had received counseling at the OPC; this proportion was the same for Van Don and Cam Pha clients (19% vs 19%; p=1.0) NOTE: At the time of the first baseline, there was no formal emotional counseling offered at either OPC, hence clients answered this question based on the degree to which they felt their emotional concerns were addressed
Clients were also asked if they had received visits at home from the home-based care team in the preceding three months Of the 816 clients who responded, 51% stated that they had received home visits Van Don clients (59%) were more likely to report home visits than Cam Pha clients (45%; p<0.0001) The 416 clients who reported home visits received a median of three home visits each (IQR: 2-3 visits); there was no difference in the number of visits reported by Van Don and Cam Pha clients (p=0.99) Respondents overwhelming reported that CHBC visits were beneficial
Almost all clients (820) responded to a question about their attendance at PLHIV support group meetings in the preceding three months; 26% reported attendance Van Don clients (38%) were more likely to attend these meetings than Cam Pha clients (18%; p<0.0001) The 216 clients who attended these meetings reported attending a median of three support groups (IQR: 2-3 meetings); there was no difference in the number of meetings attended by Van Don and Cam Pha clients (p=0.60) Clients who attended support group meetings overwhelming reported that they were helpful
Social support, stigma and discrimination and service needs
Clients reported a wide range of responses when asked about the number of people they could count on for help if they had serious problems The median was four people (IQR: 2-6 people) Van Don clients reported more people as providers of support (median 4; IQR: 3-7) than Cam Pha clients (median 3; IQR: 2-5; p<0.0001)
The median Functional Social Support Questionnaire (FSSQ) score for 808 respondents was 1.4 (IQR: 1.0-1.9), which indicates a level of support between ‘As much as I would like’ (1.0) and ‘Less than I would like’ (2.0) The median FSSQ score was 1.3 for Van Don clients (IQR: 1.0-1.7) and 1.4 for Cam Pha clients (IQR: 1.0-1.9), with no evidence of a difference between the two groups of clients (p=0.21)
Multivariate linear regression was used to explore the relationship between OPC and the FSSQ outcome scale Higher level of education was associated with better FSSQ outcomes, while having one or more children was associated with poorer outcomes Use of CHBC services, being male, higher education and higher SF-36 physical scale score were associated with better social support Of the 777 clients who responded to a question about how easy it would be to get help from their neighbors, 62% said it would be ‘very easy’ or ‘easy’ A higher proportion of Van Don clients (67%) than Cam Pha clients (58%) reported that it would be easy or very easy (p=0.01)
(28)The most commonly cited groups of people who perpetrated discrimination were neighbors (68%), family members other than parents, spouses or children (39%), the client themselves (self-stigma) (36%), people at the market (25%), and people at work (23%) The most commonly reported acts of discrimination included: gossip (66%), saying mean things (58%), staring (36%), and ‘other’ acts (38%) There were no differences between Van Don and Cam Pha clients in the proportions on the nine different types of discrimination
The most commonly reported coping strategies were: trying to distract oneself (43%), talking with family (26%), work (26%), other things (21%), talking with friends (20%), sleeping (16%) and drinking alcohol (12%) There were no differences between Van Don and Cam Pha clients in the proportions on the various coping strategies
More than 95% of respondents identified one or more worries/concerns The most commonly identified worries included: what will happen to them and their health (71%), feeling ill (69%), not having access to ART (53%), access to sufficient food (51%), pain (50%), death (43%), being alone (42%), finding a job (41%), being sent to an 05/06 center or prison (20%), and drug use (19%) The distribution of these worries was similar among Van Don and Cam Pha respondents, but Van Don clients were less likely to worry about getting married (6% vs 11%; p=0.007) and drug use (16% vs 22%; p=0.049) Among the parents, 94% were worried about their children’s future, and 87% were worried about being able to send their children to school Van Don clients were less likely to worry about their children’s future (91% vs 96%; p=0.03) and about being able to send them to school (84% vs 90%; p=0.049)
The hopes most commonly expressed were: that a cure would be found for HIV (95%), to live for a long time (89%), to be happy (87%), to be loved (85%) and to find a job (53%) Among patients who were parents, 90% looked forward to their children going to school The patterns of responses were similar in both OPCs, but Van Don clients were less likely to report looking forward to getting married (11% vs 17%, p=0.02) This, in part, reflects the fact that a higher proportion of Cam Pha clients were single, less likely to hope to have a baby (13% vs 21%; p=0.004), and less likely to hope to quit drugs (13% vs 20%; p=0.01) A larger proportion of Van Don clients expressed the hope of not being in pain (82% vs 74%); p=0.004)
Almost all clients responded to a question about the support they needed to allay their worries and help them achieve their hopes Respondents reported a median of five supportive services that would help (IQR: 4-6), with no difference between Van Don and Cam Pha clients The most commonly expressed need was a cure for HIV (95%), closely followed by care for symptoms (90%), emotional counseling (67%), and help finding a job (50%) Among those with children, 79% needed support with schooling The pattern of responses was similar across the two OPCs, but Van Don clients were more likely to report need for employment support (57% vs 46%; p=0.002) and less likely to report need for support to have a baby (7% vs 13%; p=0.01)
Drug and alcohol use and service needs
(29)Table 7: Use of drugs and alcohol, overall and by OPC*
Use of drugs or alcohol All
OPC
Van Don Cam Pha p-value*
N % % %
Ever injected illicit drugs 503 61.2 65.8 57.9 0.02 Ever smoked opiates 427 51.9 51.0 52.6 0.67 Ever injected illicit drugs or smoked
opiates 522 63.5 67.8 60.4 0.03
Smoke cigarettes 492 59.9 60.3 59.7 0.89
Consumed alcohol in the last month 382 46.8 50.1 44.4 0.09
Total 822 100% 100% 100%
-* Italicized p-value indicates overall statistically significant difference between Van Don and Cam Pha # Missing data (smoking = 1, alcohol = 6) excluded before calculation of percentage
Participants who self-reported injection drug use were asked if they had injected drugs in the last six months Of the 493 respondents, 139 (28%) reported injection in the last six months, 82 of which reported injection from or times to daily in the last week The rate of use in the past six months did not differ between Van Don and Cam Pha IDU (23% vs 30%; p=0.27)
Of the 427 clients classified as having previously smoked opiates, only 352 responded to a question about the average frequency of use in the last six months Overall, 91% reported no use in the last six months, with no difference in the pattern of use between Van Don and Cam Pha clients (p=0.61) Overall, 492 respondents were cigarette smokers (60%), with no difference between Van Don and Cam Pha clients (p=0.89) The smokers reported a median of 10 cigarettes per day (IQR: 10-20 per day) with no difference between Van Don and Cam Pha smokers (p=0.94) More than half of respondents (53%) reported never consuming alcohol, while 11% reported daily drinking There was no difference overall in the frequency of alcohol consumption between Van Don and Cam Pha clients
Most respondents (87%) reported no need for drug and alcohol services Among those that indicated a need for services, 8.4% wanted methadone services, 6.0% wanted relapse counseling, 5.5% wanted alcohol use counseling, and 4.6% wanted access to clean needles and syringes Summary
Participants in both OPCs had a median age of 31 years, and approximately two-thirds were male Van Don participants had fewer years of education and were more likely to be currently married and to have children
(30)Van Don and Cam Pha participants both reported a median of 10 symptoms in response to questions from the Memorial Symptom Assessment Scale, and showed similar profiles for all four scales (the Global Distress Index, Psychological symptoms, Physical symptoms and the Total score) Those on ART versus those not on ART reported a similar level of symptoms in all but seven of the 36 symptoms included A comparison of patient self-reported symptoms with those recorded in the medical notes for the last clinical visit found consistently higher rates (ten times higher on average) by self-report This suggests a lack of systematic health provider assessment of patient symptoms After receiving treatment, reported symptoms were fully resolved in 26% of patients About 70% of participants reported moderate or severe pain as their worst pain experience in the last month, while only 13% reported moderate or severe pain as their average pain In both cases, Van Don participants were more likely to report moderate or severe pain, but the two clinic populations did not differ in the degree to which pain interfered with daily activities of life as measured by the Brief Pain Inventory Pain was highly undermanaged by both OPCs Of the 474 PLHIV who reported moderate to severe pain in the past month, only 23 received codeine or morphine to treat their pain
At baseline, about 25% of participants were assessed as having clinical anxiety using the standard cut-off for the Hopkins Symptom Checklist, 42% were assessed as having clinical depression, and 37% of clients were assessed as having a mental health problem These proportions were similar in the two clinic populations
Van Don and Cam Pha participants did not differ in their median functional support scores as measured by the Functional Social Support Questionnaire A higher proportion of Van Don participants reported experiencing discrimination in the last three months (21% vs 16%), but the difference was not statistically significant
In relation to drug and alcohol use, the two clinic populations did not differ in the proportion of those who reported smoking opiates (ever), smoking cigarettes, or consuming alcohol in the last month Van Don clients were more likely to report having injected drugs (66% vs 58%), though there was no statistical difference between clinic populations for those who reported drug use in the past six months
5 Survival analysis in the post-intervention period
Survival analysis was undertaken to gain insight into causes of death, serious illness, and loss to follow-up (excluding transfers to other OPCs) in the post-intervention period The results are discussed below, separately for each outcome
Survival without serious illness
(31)Figure 3: Kaplan-Meier curve of survival and well enough to participate, stratified by OPC
1.00 0.98 0.96 0.94 0.92
0.90
100 200
Control Intervention 300
Days since baseline interview
Probability
400 500 600
0
The final multivariate model for survival without serious illness found that CD4 (<350 count/ mm3), male sex, older age, unemployment, and not being on ART were all associated with higher mortality and serious illness After adjustment for covariates in proportional hazards regression, the difference between OPCs was no longer statistically significant (p=0.08)
Any loss to follow-up
The endpoint of any loss to follow-up includes death (23), illness (4), discontinuation (6), loss of contact (46), and arrest (50), for a total of 129 (events) Transfer to another OPC (16 cases) was considered a censoring event Overall retention from the second baseline was 82% at 516 days; Van Don clients had a slightly higher level of retention (84%) than Cam Pha clients (81%), but the univariate comparison was not significant by proportional hazards regression (p=0.12) The Kaplan-Meier curve of retention, stratified by OPC, is shown in Figure
Figure 4: Covariate-adjusted Kaplan-Meier curve of predicted retention, by OPC
1.00 0.95 0.90 0.85 0.80 0.75 0.70
100 200 300 400 500
0
Days since baseline interview
Probability
Control Intervention
(32)Summary
Univariate analyses suggest that after the intervention was introduced, Van Don clients had higher healthy survival For death or serious illness, adjustment for covariates reduced the difference between the clinics However, there is still a strong (though statistically non-significant) suggestion that Van Don clients had lower mortality after the start of the intervention than Cam Pha clients, even after adjustment
Univariate analysis of the difference between OPCs in all losses to follow-up (excluding transfer of care to another OPC) was less pronounced than for death and serious illness alone The difference between clinics was not significant after adjustment for covariates
6 Change over time
This study seeks to assess whether an intervention at Van Don OPC led to improved health and psychosocial outcomes for its patients
At baseline, no meaningful differences were identified between the following outcomes by clinic: physical and psychological symptoms, symptom distress, symptom burden, pain prevalence, depression, anxiety, clinical mental health symptoms and pain interference However, differences were documented in one of the outcomes: worst and average pain scores and social support (Pain intensity was higher in Van Don while social support was worse in Cam Pha) The collection of multiple rounds of data over the course of the study provides for an analysis of change in outcomes over time As there were repeated measures for individuals (up to four separate rounds), mixed effect models that allow for correlation within subjects over time were used to analyze changes over time
The analysis team developed all statistical models around underlying continuous data scales and checked all assumptions (normal distribution of residuals, equal variances and appropriateness of fitted forms) For simplicity, interview rounds were used to model time, from Baseline #2 to Follow-up #1, to Follow-Follow-up #2 These models were assessed to be as good as, and sometimes better than a time model utilizing days Analysis started from Baseline #2 because the intervention at Van Don took place between this interview round and Follow-up interview #1 Starting from Baseline #2 also controlled for regression to the mean found in non-randomized control trials If there was an effect of intervention, there should be a difference in the slope of change over time between Van Don and Cam Pha clients: if the intervention is effective, Van Don client health should improve faster,
or deteriorate slower, than that of Cam Pha clients.
The team assessed each covariate for its impact on the baseline value of outcome (at Baseline interview #2), and on change over time over the three interviews Additional details of the modeling strategy are described in the Methods section The final models seek to look at the effect of OPC on change over time in outcome, adjusted for all significant covariates The key results are described below
Physical and psychological symptom outcomes
(33)Van Don, the average number of reported symptoms and the total symptoms score reduced over time, the psychological and physical subscale scores were unchanged, and the GDI subscale score increased, but at a lower rate than at Cam Pha
This study suggests that the intervention at Van Don may have resulted in reductions in the following symptom outcomes:
Number of symptoms (p=0.01)
Total symptom score (36 items; p=0.01)
Global Distress Index (distress and discomfort caused by symptom) (10 items; p=0.04)
Psychological distress and severity score (10 items; p=0.047)
There is not enough strong evidence that the intervention improved the MSAS physical score (6 items) (p=0.16)
Figure 5: Symptoms (MSAS) over time, by clinic, adjusted for covariates Number of symptoms
Rate of change p=0.02 ms Rate of change p=0.02Total score
B2 FU1
Round
FU2 Cam Pha Van Don
Number of MSAS symptoms
12
11
10
9
8
Total MSAS score
B2 FU1
Round
FU2 Cam Pha Van Don 0.8
0.7
0.6
0.5
0.4
Global distress index (GDI)
Rate of change p=0.04 Rate of change p=0.05Psychological score
B2 FU1
Round
FU2
MSAS GDI score
1.2 1.1 0.9 0.8 0.7
0.6 Cam Pha Van Don MSAG Psychological Index score 1.2 1.1 0.9 0.8 0.7
Cam Pha Van Don
B2 FU1
Round
(34)Physical score Rate of change p=0.12
MSAS Physical Index score
1
0.9
0.8
0.7
0.6
0.5
Cam Pha Van Don
B2 FU1
Round
FU2
Pain outcomes
The final models for the three pain outcomes are summarized in Annex C, Table C-2, and graphs of the covariate-adjusted effect of OPC on each of the three BPI outcomes are shown in Figure At Cam Pha, worst pain and average pain were relatively unchanged, or reduced marginally over time At Van Don, the worst and average pain outcomes clearly did reduce over time As a result, being enrolled in the intervention clinic (Van Don) was a statistically significant predictor of rate of change for average pain (p=0.04) and is a borderline predictor of change in worst pain (p=0.054) These results suggest a benefit of the intervention at Van Don Pain interference scores improved at both OPCs in a similar fashion, hence there is no evident effect of the intervention on pain interference (p=0.38)
Figure 6: Pain outcomes (BPI) over time, by clinic, adjusted for covariates Worst pain
Rate of change p=0.054 Rate of change p=0.04Usual pain
BPI - Worst Pain
5
4
3
2
1
0
B2 FU1
Round
FU2 Cam Pha Van Don
BPI - Usual Pain
5
4
3
2
1
0
B2 FU1
Round
(35)Depression and anxiety outcomes
The final models for the three HSCL outcomes are summarized in Annex C, Table C-3, and graphs of the covariate-adjusted effect of OPC on each of the three BPI outcomes are shown in Figure Depression scores reduced markedly compared to anxiety scores at Cam Pha, and more so at Van Don There was a small reduction in anxiety scores over time at both clinics There was no difference in the rate of change in anxiety scores between the two clinics (10 items; p=0.45) However, depression scores (15 items) reduced sharply in Van Don compared to Cam Pha (p=0.02) The overall mental health problem score (the average across all 25 questions and an intermediate between its two components: anxiety and depression), showed a difference in the rate of change which approached, but did not reach, statistical significance (p=0.07)
Figure 7: Mental health outcomes (HSCL) over time, by clinic, adjusted for covariates Anxiety score
Rate of change p=0.45 Rate of change p=0.45Anxiety score
Mental health problem score Rate of change p=0.45
HSCL - Anxiety score
2
1.8
1.6
1.4
1.2
1
B2 FU1
Round
FU2 Cam Pha Van Don
HSCL - Depression score
2
1.8
1.6
1.4
1.2
1
B2 FU1
Round
FU2
Cam Pha Van Don
HSCL - Mental Health problem score
2
1.8
1.6
1.4
1.2
1
B2 FU1
Round
FU2 Cam Pha Van Don
Social support outcomes
(36)Figure 8: Social support score (FSSQ) over time, by clinic, adjusted for covariates Social support score
Rate of change p=0.52
HSCL - Social support score
2
1.8
1.6
1.4
1.2
1
B2 FU1
Round
FU2 Cam Pha Van Don
Summary
Overall, the models provide evidence that suggests that the intervention at Van Don resulted in reductions in the number of manifest symptoms, and improvements in the three scales measured by the MSAS (GDI, Total score and Psychological symptoms), worst and average pain (BPI), and depression (HSCL) as compared to PLHIV enrolled in Cam Pha clinic
The results not provide compelling evidence of change in pain interference in daily life activities (BPI), anxiety (HSCL) or functional social support (FSSQ) The results are less clear in relation to physical symptoms (MSAS), where the pattern is similar to the other MSAS scales, but is not statistically significant, or in overall mental health (HSCL), which is a mixture of the anxiety and depression scores
7 PLHiv and health care worker perceptions of the intervention
A total of 35 PLHIV and 14 health care workers were interviewed at Round to assess initial feedback on the intervention in Van Don The interviews comprised of two focus groups (one male, one female) with PLHIV leaders at the clinic, 20 in-depth interviews with PLHIV clients who were identified as patients with ‘typical’ palliative care problems, and 14 in-depth interviews with health care workers from the Van Don OPC, and health leaders from the District Hospital where the OPC is based [NOTE: Words in bold represent emphasis of passage by author]
Acceptability of the symptom assessment tool
The doctors in Van Don were asked to provide their opinions on the brief symptom-screening tool that was introduced as part of the intervention to improve symptom detection and treatment In general, they described the screening assessment questions as helpful:
(37)I think it is much better Now we care for clients more - not just their health, I mean, physical health - but [we] also ask more about feelings, pain etc It is helpful that we have a set of questions to follow Without it, we tend to forget to ask those questions [smiling] – Doctor,
VD OPC
As evident in these statements, the doctors also acknowledged the more ad hoc way in which they assessed symptoms previous to the introduction of the symptom-screening tool
To triangulate these findings, patients included in the qualitative interviews were also asked how the OPC doctors assessed them on their monthly clinic visits All patients interviewed reported that they had been asked to list and scale the severity of their symptoms
[The doctors] asked me how bad the headache was, where I had pain, and what caused my pain They asked me everything – Male PLHIV, FGD VD
Yes, they asked what level my pain was using a 10-point scale – Male PLHIV, FGD VD The doctors asked me to score the severity of my pain from level to 10 I answered that my pain was at level or Before, when I had less pain, it was at level 3… – Female PLHIV, VD
Another patient verified when the use of the scales had begun:
Use of that scale started just recently, about three to four months ago Yes, they [OPC staff] have used it with me recently – Female PLHIV, VD
Acceptability in relation to health care worker and patient interactions
The qualitative interviews indicated that OPC health care workers in Van Don were consistently positive, and surprised even, by the measure to which interactions with patients had improved
…I think patients this year feel much more comfortable with me They answer with ease whenever I ask them questions They are also not hesitant to ask me questions In general, it is quite a good atmosphere in the clinic and my consultation room now – Doctor, VD OPC I am quite confident with my job now [smiling] I understand patients better They also know me better – Doctor, VD OPC
After the training, we have changed our working style Of course, for the better [smiling] We have become more service- and result-oriented In addition, we now know how to listen to our clients and encourage them to share their thoughts and feelings with us – Mental Health
Team member, VD OPC
Generally, I think patients seem more comfortable….and… pleased They are more relaxed… have more trust in me, talk to me more Personally, I think they must feel happier receiving more attention from the staff here, even if they provide such feedback to us – Doctor, VD OPC
I provide counseling to patients more carefully Since this program started, my working method has changed Yes, it’s better Patients are also friendlier [smiling] – Pharmacy Chief,
(38)I think the doctors in the OPC are very dedicated Yes, there have been a lot of changes When we come to the OPC, health staff are friendlier and ask more questions than before – Female
PLHIV, VD
In the past, the health staff in this clinic didn’t know much about HIV so we [PLHIV] didn’t have close relationship with them Now they have the required knowledge and are more open and considerate – Male PLHIV, FGD VD
The services are getting better and better every passing day – Female PLHIV, FGD VD
A few patients described ease in asking for symptom care and treatment in the OPC
[The intervention] has been very helpful because now I don’t have to buy medicine from outside When I feel ill or have pain, such as headache or stomachache, I can ask for medicine immediately – Male PLHIV, VD
Acceptability of treating pain with amitriptyline and oral morphine
The Van Don OPC health care team readily adopted amitriptyline for the treatment of peripheral neuropathy and depression Use of oral morphine in Van Don was more problematic; physicians demonstrated substantial reluctance to its use for PLHIV clients Conversely, client acceptability was substantially high Those interviewed generally agreed on the positive value of using morphine to treat pain
I have never taken morphine here However, if I have severe pain, and the doctor said that I need that medicine, it would be fine, as long as it helps release pain
– Male PLHIV, VD
I am sure I will take it immediately if I am in a severe pain! I only want to get rid of it [the pain] as soon as possible, so I am sure I will take it right away – Male PLHIV, VD
At first I thought that there would be a bad effect if we inject morphine Now, after thinking it through…yes, if I were in pain, there wouldn’t be any problem using it as a painkiller – Male
PLHIV, VD
I think there is no problem if we use it in accordance with the doctor’s instructions I am not afraid of addiction As far as I know, we cannot buy morphine as we can a bundle of vegetables from the market There must be a prescription from the doctor and then we are allowed to use it – Male PLHIV, VD
The above quotes indicate acceptability based on reasoning Those interviewed suggested that they would accept morphine for pain because it works, because they believe that pain treatment outweighs the negative effects of morphine, and because, if prescribed by a doctor, its use would be legal and acceptable
Acceptability of mental health services
(39)Initially, we did not think that clients would use such a service We did not think that the service would be successful However, within [just] three months, 100 clients accepted the service Quite amazing – Doctor, VD OPC
One of the mental health counselors described the patients as nervous the first time they came for counseling, but that this dissipated relatively quickly His background as a former drug user also enabled him to speak with patients freely about the realities of drug use
People with HIV have very low self-esteem They are not confident, [and] are always blaming themselves Clients referred to this room were generally quite nervous Once they realized they had someone to talk to, someone to share with, they immediately felt more comfortable With drug users, it was a bit more difficult Being a former drug user, I know how it felt So the way I counseled was first just to agree with them that yes, it was such a pleasure to use drugs, and that it was extremely difficult to refrain from drugs No one ever told them that They would then feel that they could trust me They felt relaxed – Mental Health Team Member,
VD OPC
Patients that sought support from the mental health team described their experience positively
The most helpful service is medicine [ART], second is Duong Sinh Tu Na, third is psychosocial support – Male PLHIV, VD (AIDSTAR 2010)
Sometimes I come to talk with the mental team, whenever I’m sad Any month that I’m sad and have some problems with my mind, I also come to share, to talk with them! – Male
PLHIV, VD
[The mental health team] helps me, for example, when I have difficulty sleeping – Male
PLHIV, VD
In May 2010, during an AIDSTAR initiative to document the mental health components of the palliative care intervention in Van Don, patients were asked how mental health services had impacted their lives (if at all) Patients spoke about the skills and confidence they acquired that altered the way they experienced and handled problems
I still have problems, but now I am handling them differently – Male PLHIV, VD Before, my bones felt very tired Now, with massage and exercise, I feel more relaxed, confident, and don’t worry as much – Female PLHIV, VD
I feel confident and more relaxed I don’t worry as much anymore and I sleep better – Male
PLHIV, VD Feasibility
In response to the question on whether the intervention could be replicated and implemented in other HIV clinics, some Van Don staff expressed confidence that it could be applied with ease
(40)To me, it is not difficult at all Our patients can benefit from this integration So I think it is feasible to integrate palliative care into existing services provided by the clinic – Mental
Health Team member, FGD VD
In my opinion, we should apply palliative care in all clinics PLHIV everywhere have the same needs I know in Cam Pha there are only CHBC teams, but not palliative care/mental health teams Palliative care is only applied in Van Don – Mental Health Team member, FGD VD
The head of the OPC in Van Don offered a more nuanced opinion by stating that the intervention could only be replicated where the clinic staff were dedicated and “passionate” about their work
I think it is neither easy nor difficult But palliative care…is very necessary indeed We need to be very passionate with our job and patients to this well Without the passion, you can’t work well, even if you have been working for ten years with a lot of experience And vice versa: if you love your job, you can it well quickly in the first few months – OPC Chief, VD OPC
Similarly, the head of the hospital’s Infectious Disease and Inpatient Department felt that if health care workers were compassionate, then integration of palliative care could be achieved
I think that we can many things if we have a good heart It’s not difficult if we have a sense of mercy for the people around us Like me, I am a doctor and I have to provide care to 80, sometimes even 120 patients a day It’s a heavy job, but I still have time to talk with them, to counsel them That’s why I think it’s not too difficult You are right when you mentioned that some health staff are not that committed, but I think more than half of our staff [work] whole-heartedly Yes, we have some people…’one rotten apple spoils the basket’, and we can’t avoid that However we need to help them increase their awareness so that they are closer to patients and understand that we need to treat both: their physical and mental problems! –
Hospital Infectious Disease and Inpatient Department Chief, VD
8 Limitations
Quasi-experimental studies can present challenges to internal and external validity This study applied a strong design for a non-randomized control trial that eliminated all but three threats to internal validity: 1) selection maturation, 2) instrumentation, and 3) local history
In this study, two clinics with observably similar PLHIV populations in adjacent districts were compared to each other in an effort to minimize differences between the two comparative populations Given the inherent non-equivalent nature of the two groups, there were measurable similarities and differences in the two clinic populations at baseline Variables including education, having a family and children, time of clinic enrollment, ART status, and levels of pain intensity were significantly different between the two clinics Multiple regression analysis adjusted for baseline covariates, controlling for potential confounding However, this did not rule-out ‘unobserved’ variables (those that were not measured but could have been important factors to control for), and their potential impact on outcome findings
(41)When two groups are non-equivalent, there is also a possibility for selection-maturation, where variables in the control or intervention group change at different rates (e.g becoming sicker, or less sick over time) The use of scales can also lead to exaggerated scores in quasi-experimental studies, resulting in inflated positive or negative findings A double baseline was used to control for these threats No secular maturation trends were observed between the two baselines To eliminate the potential threat of regression to the mean, only the second baseline was used in the analysis In some cases, a significant external event in the control or intervention site (for example a complete change in staff in a clinic, or switching from free to at-cost services) may have been a contributing or sole cause of positive or negative change in the intervention or control site The use of qualitative data to track alternations that might have occurred during the study, can help assess the impact of these external phenomena Other than the pause in the study during April/May 2009 that occurred for both study sites before the intervention was initiated, no other recorded event that could have influenced outcomes was identified
External validity (or generalization) is the degree to which results from one study can be extrapolated and applied to other people and settings In Viet Nam, typical HIV clinic clients are characterized as primary male (from 70-80%), where the majority of men were infected through heroin injecting drug use, and women through their IDU partners (Huong et al 2011; Klotz et al 2007; Tran et al 2010) [19, 20, 21] The average age of patients cited in these papers was 27-35; and 70-90% of the HIV clinic patient population was on ART This pattern is found in provinces where HIV transmission is primarily due to IDU behavior, rather than sexual transmission (MoH 2010) [23] These types of provinces, including Quang Ninh (where the study was conducted), Hai Phong, Hanoi and Ho Chi Minh City in the south, comprise of the majority of HIV infections in the country The study population in Cam Pha and Van Don were similar to this general profile of PLHIV in Viet Nam This indicates an affinity between the study population and those described in other recent publications
The Viet Nam MoH has opted for a decentralized district-focused model of HIV care The majority of OPCs are based at the district government hospital and draw on hospital resources such as pharmacy, laboratory and infectious disease department staff to support the functioning of the clinic The OPCs in Cam Pha and Van Don are similar in structure and service delivery to those run with support from PEPFAR, of which there are more than 50 clinics
These factors suggest that results from the palliative care integration study could be replicated in other similar HIV outpatient care settings in Viet Nam To what degree they are applicable outside this spectrum is unknown
9 Conclusion and recommendations
This study compared standard care in one OPC with a multifaceted intervention at another to improve patient health and psychosocial outcomes by doing the following:
1 Improving HIV provider skills in palliative care and enhancing their belief in its importance Establishing systematic palliative care assessment/screening and care
3 Increasing access to and use of essential palliative care medication
(42)The data show that the intervention at Van Don OPC was followed by improvement in a number of important palliative care outcomes After adjustment for covariates, the following changes were noted following the intervention:
Symptom number and frequency, severity, and distress, as measured by a number of scales, increased over time in Cam Pha, but decreased slightly, stayed the same, or increased at a slower rate at Van Don
Worst pain and usual pain over the last month reduced slightly in Cam Pha, and reduced more sharply in Van Don
Depression and overall mental health scores improved slightly in Cam Pha, versus sharp improvement in Van Don
Mortality and serious illness, while relatively rare, appeared lower in Van Don than Cam Pha, but this difference was not statistically significant (p=0.08)
Palliative care outcomes that did not improve following the intervention included interference of pain in daily life, anxiety, and functional social support In addition, no difference was found in overall loss to follow-up
Recommendations
This study provides important insights into under-addressed physical and psychological care needs of PLHIV It also suggests concrete steps that can be made to address these needs The following are specific recommendations to be taken under consideration in future PLHIV OPC care 1 Depression and anxiety
Rates of clinical depression and anxiety are very high in this population and have been observed as elevated in other recent studies in Viet Nam[19] In addition, the majority of PLHIV with anxiety in this study were also depressed, so it is important to address these two mental health problems together Depression in PLHIV is associated with significant levels of sub-optimal ART adherence, sexual risk and disease progression Methadone expansion will likely help to reduce psychiatric morbidity for those who are eligible However, for those who are no longer active injectors, it will remain important Studies in Viet Nam, including the one in Quang Ninh, indicate that female PLHIV are at significant risk of depression and less social support than men [19, 20] Therefore, the following are recommended:
Screening and assessment: Depression and anxiety screening of OPC clients should be conducted at least twice a year Screening can be done by the OPC doctor, adherence counselor, CHBC team, or OPC treatment supporter The initial screening can involve a simple two-question tool about presence and severity of sadness and anxiety (see example in Annex F) Those with mental health problems could then be referred to a counselor, or CHBC team for more detailed assessment The Hospital Anxiety and Depression Scale is a short 15-question tool that helps providers to identify whether someone has moderate or severe depression and/or anxiety This, or another simple and rapid diagnostic tool, is recommended for use in OPCs
Counseling and support: Basic counseling and emphatic communication should be provided to all clients with depression and/or anxiety In addition, practices such as Duong Sinh can be taught to, and led by, PLHIV so they are able to manage painful emotions and life situations
(43)2 Pain and other symptoms
Pain and physical symptoms were found to be common and prevalent in the majority of PLHIV (98.4%) in this study, both those on ART and not on ART Patients reported a median of ten symptoms in the past month In comparing OPC patient records to PLHIV symptom self-reports, the records showed fewer than 10% of symptoms This suggests that there are significant gaps in the degree to which OPC doctors routinely assess for symptoms PLHIV with symptoms also identified a high degree of under-treatment of their symptoms This was particularly striking for PLHIV with severe pain, few of whom received appropriate treatment Symptoms are associated with poor ART adherence and disease progression, but studies indicate that where symptom treatment improves, poor ART adherence reduces [21, 22, 23, 24] To improve detection and treatment of symptoms, the following are recommended:
Screening and assessment: A short, one-page screening tool (e.g one that includes the ten most common symptoms in PLHIV) should be used by OPC health care workers (or by a PLHIV clinic volunteer where health care workers are too busy) during routine patient visits This tool should allow patients to rate the severity of their symptoms on a scale on 1-10 Thus, OPC health care workers will be able to track changes in severity over time
Treatment of pain: Symptomatic reports indicating peripheral neuropathy were high in this study Peripheral neuropathy was almost entirely undetected and untreated in patients This phenomenon appears common in OPCs Amitriptyline is very effective in treating peripheral neuropathy and is available in a number of OPCs across the country OPC health care workers need to assess for peripheral neuropathy on every clinic visit since this debilitating form of pain can lead to permanent nerve damage and impaired mobility Codeine is available in PEPFAR-supported clinics and should be used to treat moderate pain There are five HIV OPCs in Viet Nam that have procured oral morphine and are treating severe pain in PLHIV It is recommended that OPCs make oral morphine available so that severe pain does not go untreated This is essential for a minimum standard of care
Treatment of other symptoms: OPCs need to treat common and distressing symptoms such as insomnia, cough, and nausea proactively This involves improved utilization of medicines made available at OPCs, and ensuring referral to CHBC teams for follow-up care and monitoring 3 Future research
(44)AnnEx
Annex A: methods for multivariate model
Cross-sectional multivariate models at baseline
All continuous models included a variable for OPC, and were modeled using multivariate linear regression techniques with the following procedures:
1 Potential confounders identified a priori were added to clinic as forced variables in the preliminary main effects model (see Table A-1 for a listing of these variables)
2 A selection of other potential confounders were screened by univariate linear regression, and added to the preliminary main effects model if found to be associated with outcome with p<0.2 The variables screened are specified in Table A-1
3 Potential confounders added to the preliminary main effects model were examined, and those that were not statistically significant predictors of outcome (p≤0.05) were removed one-by-one, starting with the least statistically significant until all the remaining potential confounders identified were significant at p≤0.05 Statistical significance was assessed by the t-test for continuous or dichotomous categorical variables, and by the partial F-test for polychotomous categorical variables modeled as dummy variables
4 First-order interactions of clinic by each of the main effects models were created by multiplying the corresponding variables, and added one-by-one to the main effects model If the interaction term was significant at p≤0.05, it was added to the preliminary final model First-order interactions not including clinic were not considered
5 The pentultimate model was run including all interactions terms which when added singly had shown statistical significance at p≤0.05 Any interaction term no longer found to be statistically significant at p≤0.05 was removed, starting with the least statistically significant The final model thus included: clinic, all potential confounders identified a priori, other
potential confounders which remained statistically significant at p≤0.05 in the main effects model, and first-order interaction terms, including clinic, which remained statistically significant at p≤0.05 in the final model
7 A histogram of the jackknife residuals and QQ plots were used to examine the appropriateness of the model If clearly non-normal, the outcome variable was transformed using a natural log transformation and fit was re-assessed As many of the outcomes were scales with possible zero values, log transformations were performed after adding a constant to the scale value If fit was substantially improved by the transformation, the revised model was used
Dichotomous outcomes (clinical anxiety, clinical depression, or poor mental health) were analyzed using multivariate logistic regression, using the same risk factors and modeling strategy described above Overall goodness of fit was assessed by examining the Hosmer-Lemeshow Goodness of Fit test, and the area under the ROC curve
Survival analysis
(45)Table A-1: Variables controlled for in multivariate analysis, and whether specified a priori (forced), or screened for
Variable (if two or more, Specification
select best):
Cross-sectional
(Linear or Logistic)
Survival
Analysis Mixed Effects
Sex Male/Female A priori Screen Screen
Age Continuous A priori Screen Screen
Employment
Full-time/Part-
time/Unem-ployed A priori Screen Screen
Employed/Not
Living alone Yes/No A priori Screen Screen
CD4 count
Continuous
A priori Screen Screen
<200 / 200+ <350 / 350+ ART adherence
Not on ART/ 95%+ / <95%
A priori Screen Screen
On ART/ not on ART
SF36-Physical Continuous A priori Screen Screen
Discrimination Yes/No A priori Screen Screen
Illicit drug use
IDU – ever use
A priori Screen Screen
Opioid smoking - ever
IDU/opioid - ever Recent IDU (6 months) Recent opioid smoking (6 months) Detention in prison or 05/06 center
Alcohol Any in last month / None A priori Screen Screen
Recent smoking Any / None A priori Screen Screen
Education <6 / 6-9 / 10-12 / Tertiary Screen Screen Screen
Marital status
Married / single /
separated Screen Screen Screen
Married / not
Children Yes/No Screen Screen Screen
(46)Change over time
Change was assessed in rounds 2-4 which allowed linear models that compared change in slope in the clinics No risk factors were specified a priori, so all variables in Table A-1 were screened before inclusion in the models, with simultaneous univariate screening for effect on intercept and change over time No interaction terms were considered Time was modeled as round, or as time in days from the Round interview In the absence of evidence of the superiority of time in days, models using time as interview round were preferred as this matched the roll out of intervention, which was progressively implemented between Round and Round interviews
The appropriateness of the models was assessed by the following:
Assessment of whether data were better modeled as raw (highly skewed in univariate analysis) or after transformation (log or square root)
Examination of the normality of the distribution of residuals (within and between individual) using QQ plots of standardized residuals, and examination of scatter plots
Examination of the distribution of residuals over time to assess homoscedasticity
Annex B: Cross-sectional multivariate model results
The tables below show the results of the cross-sectional multivariate models at Baseline #1 The following codes are used in tables B-1 and B-2:
= Significant independent predictor (no interaction) with higher outcome scores (worse outcomes) for participants with higher values of this predictor
= Significant independent predictor (no interaction) with lower outcome scores (better outcomes) for participants with higher values of this predictor
X = A significant interaction with OPC - = Not a significant predictor
Table B-1: Final cross-sectional models for five symptom outcomes (MSAS)
Covariates Number of Symptoms
symptoms Total score GDI Psychological Physical
Van Don X X X X X
Male
Age (18-48) X[a] X[a] X[a] X[a]
-Not working (vs full time)* - - - -
-Part-time (vs full time)*
-Living alone - - - -
-CD4 (0-1500) X[b] X[b] X[b] - X[b]
Not on ART (vs ≥95%
adher-ent)* -
<95% adherent (vs ≥95%)* - - - -
-SF-36 physical (0-100)
(47)-PLHIV support group user - - - -
-Emotional counseling - X[c] X[c] X[c]
-Discriminated against X[d] X[d]
Illicit drug user#
Recent alcohol user -
-Smoker - - - -
-% of variance explained (R2) 34% 41% 36% 29% 37%
* Categorical variables with three levels (work status and ART and adherence status) only shown if the overall variable is significant AND the individual comparison is significant
# Modeled as use of injection drugs in last months in all models.
[a] Age X OPC: At Van Don, older clients have worse outcomes than younger clients; the opposite is true at Cam Pha [b] CD4 X OPC: Higher baseline CD4 counts are associated with worse outcomes at Van Don, and better outcomes at Cam Pha [c] Emotional counseling X OPC: Van Don clients who reported receiving emotional counseling had worse outcomes than those not receiving emotional counseling; there is no difference between the groups at Cam Pha
[d] Discrimination X OPC: Cam Pha clients who reported discrimination had much worse outcomes than those who did not report discrimination; at Van Don the results are in the same direction, but less extreme
Table B-2: Final cross-sectional models for three pain outcomes (BPI)
Covariates Pain
Worst Average Pain interference@
Van Don X X
Male
-Age (18-48) - -
Not working (vs full time)* - -
-Part-time (vs full time)* - -
-Living alone - -
-CD4 (0-1500) - -
-Not on ART (vs ≥95%
adherent)* -
-<95% adherent (vs ≥95%)* - -
-SF-36 physical (0-100)
CHBC service user X[a] - X[a]
PLHIV support group user - -
Emotional counseling - -
-Discriminated against
Illicit drug user# X[b]
-Recent alcohol user - -
-Smoker - -
-Percentage of variance
explained (R2) 18% 20% 12%
* Categorical variables with three levels (work status and ART and adherence status) only shown if the overall variable is significant AND the individual comparison is significant
# Recent smoking of opiods (Worst), ever use of injection drugs (Usual) and recent injection drug use (Interference).
(48)Table B-3 presents final logistic regression models and uses a similar coding, but with a different interpretation
= Significant independent risk predictor for outcome (no interaction)
= Significant protective predictor for outcome (no interaction)
X = A significant interaction with OPC
- = No statistically significant risk or protective effect
Table B-3: Final cross-sectional models for three mental health outcomes (HSCL)
Covariates Mental health outcomes
Anxiety Depression Poor mental health
Van Don X X X
Male -
Age (18-48) X[a] X[a]
-Not working (vs full time)* - -
-Part-time (vs full time)* - -
-Living alone - -
-CD4 (0-1500) - -
-Not on ART (vs ≥95% adherent)* - -
-<95% adherent (vs ≥95%)* - -
-SF-36 physical (0-100)
CHBC service user - -
-PLHIV support group user - -
-Emotional counseling - -
-Discriminated against
Illicit drug user# X[b]
Recent alcohol user - -
-Smoker - -
-Area under the ROC curve 0.76 0.79 0.79
* Categorical variables with three levels (work status and ART and adherence status) only shown if the overall variable is significant AND the individual comparison is significant
# Recent injection drug use (all three outcomes)
[a] Age X OPC: At Van Don, higher age is a risk factor for outcome, while at Cam Pha it is not
(49)Annex C: Covariates included in longitudinal models
The tables below show the variables adjusted for in the mixed effect multivariate models over the period Baseline interview #2 to Follow-up interview #2 Variables not mentioned in the tables were not included in a final model for any of the outcomes for that outcome-set
Table C-1: Final cross-sectional models for five symptom outcomes (MSAS) Covariates
Symptoms
Number of
symptoms Total score GDI Psychological Physical
I C I C I C I C I C
Van Don • • • • •
Male • • • • •
Age (18-48) • • •
CD4 (0-1500) • • • • • •
Not on ART • • •
SF-36 physical (0-100) • • • • •
CHBC service user •
Emotional counseling •
Discriminated against • • • • •
Illicit drug user* • • • • • •
Recent alcohol user • •
Smoker •
Parent •
I = Intercept (baseline value at baseline interview #2); C = Rate of change (from baseline #2 to follow-up #2)
* Modeled as previous prison or 05/06 center experience in all models (except intercept in the number of symptoms)
Table C-2: Final cross-sectional models for three pain outcomes (BPI) Covariates
Pain
Worst Average* Pain Interference
I C I C I C
Van Don •
Male • • •
Age (18-48) •
CD4 (0-1500) • •
Not on ART • • •
SF-36 physical (0-100) • • • •
Discriminated against • • • •
Recent IDU •
Recent alcohol user • •
I = Intercept (baseline value at baseline interview #2); C = Rate of change (from baseline #2 to follow-up #2)
(50)Table C-3: Final cross-sectional models for three mental health outcomes (HSCL*) Covariates
Pain
Anxiety score Depression score Total clinical score
I C I C I C
Van Don •
Male • • •
Age (18-48) • • •
CD4 (0-1500) • • •
SF-36 physical (0-100) • • • • •
Discriminated against • • •
Illicit drug user# • • • •
I = Intercept (baseline value at baseline interview #2); C = Rate of change (from baseline #2 to follow-up #2) * Modeled as continuous score values, not as dichotomous categories divided at a score of 1.75
# The rate of change modeled previous prison or 05/06 center experience in all models, and the intercept in the depression model was modeled as ever use of injection drugs
Table C-4: Final cross-sectional model for the functional support (FSSQ) Covariates
Functional Support
Worst
I C
Van Don •
Male •
Living alone •
SF-36 physical (0-100) •
Discriminated against •
Ever in prison or 05/06 Center •
Education •
(51)Annex D: Symptom screening tool
Clinical symptom assessment No pain (including
backache, headache) 10 Worst pain imaginable
No cough 10 Worst cough imaginable
No fatigue/tiredness 10 Worst fatigue/tiredness imaginable No difficulty sleeping 10 Worst difficulty sleeping imaginable NoNumbness, tingling or
shooting pain (in legg/
feet or hands/arms) 10 Worst numbness, tingling or shooting pain imaginable
No lack of appetite 10 Worst lack of appetite
No nausea/vomiting 10 Worst nausea/vomiting imaginable
No itchy skin 10 Worst itchy skin imaginable
No diarrhea 10 Worst diarrhea imaginable
No sadness/depression 10 Worst sadness/depression imaginable
(52)Annex E: Pain management poster
Pain assessment management&
Time: When did the pain start?
Location: Where is the pain located? Does the pain radiate? Duration of pain: When experiencing pain, how long does
it generally last?
Intervals: Is the pain constant or intermittent?
Circumstance and place: Where were you and what were you doing when the pain started? Does the intensity of pain increase at any point during the day/evening?
Intensity of Pain: What is the intensity of your pain on the 10-point pain scale? During rest and while in motion? (See scale and instructions below)
Characteristic of the pain: How does the pain feel? Shooting, stabbing, cramping, burning…
Mediating factors: What make the pain better or worse? Treatment: Have you done anything to reduce the pain?
What medications? Did they help?
Patient’s perspective: In your opinion, what is the cause of this pain?
Are there other factors influencing pain (eg psychosocial)?
Pain management
Prescribe analgesics according to the WHO 3-step treatment ladder
Counsel the client and their family in how to use the prescribed medicines
VIETNAMESE AND AMERICANS IN PARTNERSHIP TO FIGHT HIV/AIDS 0
No pain 1 Mild pain2 3 4 Moderate pain5 6 7 Severe pain8 9 Worst pain imaginable10
On a scale from to 10, where means no pain and 10 means the worst pain imaginable, where would you score your pain right now? What is the highest score and the lowest score it has been in the past week?
Pain Intensity Scale
Provide information to the client and family on non-pharmacological therapy such as massage, acupuncture, heat therapy, cold therapy In complex cases where further support is needed,
seek guidance or refer the client for further care
Questions for pain assessment
If the client is a child, or does not respond well to the 10-point Pain Intensity Scale, use the Wong Baker FACES Pain Rating Scale for pain Categorize pain intensity as follows: =
no pain; 1- = mild pain; - = moderate pain; – = severe pain; 10 = worst pain imaginable Moderate pain Weak opioid, +/- non-opioid, +/- adjuvant Severe pain Mild pain Persistent or increased pain Persistent or increased pain Non-opioid, +/- adjuvant Strong opioid, +/- non-opioid, +/- adjuvant
If the client is unable to use either scale, ask them to grade their pain from the following choices: no pain - mild pain - moderate pain - severe pain - worst pain imaginable
This poster was supported by U2G PS001172 from CDC
Its contents are solely the responsibility of the authors and not necessarily represent the official views of CDC
(53)Annex F: Depression and anxiety screening tool
Mental Health Screening Form
1 How are you feeling lately? Is there anything making you sad? How sad? 2 Is there anything making you worried? How worried?
No sadness/depression 10 Worst sadness/ depression imaginable
No worry/anxiety 10 Worst worry/anxiety imaginable
*Refer all patients scoring or more to a counselor for further assessment
(54)rEFErEnCES
1 Krakauer, E., et al., Vietnam: Integrating palliative care into HIV/AIDS and cancer care Journal of Pain and Symptom Management, 2007 33(5): p 578-583
2 Brown, S., et al., Palliative care development: The Nepal Model Journal of Pain and Symptom Management 2007 33(5): p 573-577
3 Spencer, M., Pain relief in Thailand, in Pain and Palliative care in the Developing World and
Marginalized Populations: A Global Challenge, M.R Rajagopal, D Mazza, and A.G Lipman,
Editors 2003, Hawthorn Medical Press: New York
4 Chang, V.T., et al., The memorial symptom assessment scale short form (MSAS-SF) Cancer, 2000 89(5): p 1162-71
5 Portenoy, R.K., et al., The Memorial Symptom Assessment Scale: an instrument for the evaluation
of symptom prevalence, characteristics and distress European Journal of Cancer, 1994 30A(9):
p 1326-36
6 Breitbart, W., et al., Pain in ambulatory AIDS patients I: Pain characteristics and medical
correlates Pain, 1996 68(2-3): p 315-21.
7 Harding, R., et al., Is antiretroviral therapy associated with symptom prevalence and burden? International Journal of STD & AIDS, 2006 17(6): p 400-5
8 Wakeham, K., et al., Symptom burden in HIV-infected adults at time of HIV diagnosis in rural
Uganda Journal of Palliative Medicine, 2010 13(4): p 375-80.
9 Lee, K.A., et al., Symptom experience in HIV-infected adults: a function of demographic and
clinical characteristics Journal of Pain & Symptom Management, 2009 38(6): p 882-93.
10 Cleeland, C.S., The Brief Pain Inventory: User Guide 2009, The University of Texas: Houston, Texas
11 Caraceni, A and A Caraceni, Evaluation and assessment of cancer pain and cancer pain
treatment Acta Anaesthesiologica Scandinavica, 2001 45(9): p 1067-75.
12 Keller, S., et al., Validity of the brief pain inventory for use in documenting the outcomes of
patients with noncancer pain Clinical Journal of Pain, 2004 20(5): p 309-18.
13 Cleeland, C.S., et al., Pain assessment: global use of the Brief Pain Inventory Annals of the Academy of Medicine, Singapore, 1994 23(2): p 129-38
14 Parloff, M.B., H.C Kelman, and J.D Frank, Comfort, effectiveness, and self-awareness as criteria
for improvement in psychotherapy American Journal of Psychiatry, 1954 3: p 343-351.
15 Mollica, R.F., et al., Indochinese versions of the Hopkins Symptom Checklist-25: a screening
instrument for the psychiatric care of refugees American Journal of Psychiatry, 1987 144(4):
p 497-500
16 Antelman, G., et al., Depressive symptoms increase risk of HIV disease progression and mortality
among women in Tanzania Journal of Acquired Immune Deficiency Syndromes: JAIDS,
2007 44(4): p 470-7
17 Hopkins Symptoms Checklist-25 (Vietnamese version), Harvard School of Public Health: Boston
18 Broadhead, W.E., et al., The Duke-UNC Functional Social Support Questionnaire Measurement
of social support in family medicine patients Medical Care, 1988 26(7): p 709-23.
(55)20 Huong DT, Bannister W, Phong PT, Kirk O, Peters L Factors associated with HIV-1 virological failure in an outpatient clinic for HIV-infected people in Haiphong, Viet Nam Int J STD AIDS 2011 22(11):659-64
21 Tran VH, Nguyen NTT, Arnalda G, Burdon R, Green K, Mills S Program Evaluation of Patients
Initiating Anti-retroviral Therapy in Two Ho Chi Minh City Facilities, Hanoi: FHI, March 2010.
22 Duchesne D, Franchi O, Thi PDT, Trà MN, Brunel V, Luhmann N, Lacombe K, Lemoine M Mental health distress among HIV-infected male drug users and female sex partners in Vietnam 6th IAS on HIV Pathogenesis, Prevention and Treatment, 17-20 July, 2011, Rome, Italy
23 Ammassari A, Murri A, Pezzotti P, Trotta MP, Ravasio L, De Longis P, Lo Caputo S, Narciso P, Pauluzzi S, Carosi G, Nappa S, Piano P, Izzo CM, Lichtner M, Rezza G, Monforte A, Ippolito G, d’Arminio MM, Wu AW, Antinori A Self-reported Symptoms and Medication Side Effects Influence Adherence to Highly Active Retroviral Therapy in Persons with HIV Infection
Journal of Acquired Immune Deficiency Syndrome 2001; 28(5): 445-449
24 Chiou PY, Kuo BIT, Lee MB, Chen YM, Chuang P, Lin LC A programme of symptom management for improving quality of life and drug adherence in AIDS/HIV patients Issues and Innovations
in Nursing Practice 2006; 55, 167-179.
25 Holzemer WL HIV and AIDS: The symptom experience American Journal of Nursing 2002; 102:48-52
(56)(57)(58)