True outcomes for patients on antiretroviral therapy who are “lost to follow-up” in Malawi

Abstract

Problem

In many resource-poor countries that are scaling up antiretroviral therapy (ART), 5–25% of patients are reported as “lost to follow-up”. This figure is 9% in Malawi. There is no published information about the true outcome status of these patients.

Approach

In four facilities in northern Malawi, ART registers and master cards were used to identify patients who had not attended the facility for 3 months or more and were thus registered as “lost to follow-up”. Clinic staff attempted to trace these patients and ascertain their true outcome status.

Local setting

Of 253 patients identified as “lost to follow-up”, 127 (50%) were dead, 58% of these having died within 3 months of their last clinic visit. Of the 58 patients (23%) found to be alive, 21 were still receiving ART and 37 had stopped treatment (high transport costs being the main reason for 13 patients). Sixty-eight patients (27%) could not be traced, most commonly because of an incorrect address in the register. Fewer patients were alive and more patients could not be traced from the central hospital compared with the peripheral hospitals.

Relevant changes

Better documentation of patients’ addresses and prompt follow-up of patients who are late for their appointments are required.

Lessons learned

ART clinics in resource-poor countries should ensure that patients’ addresses are correct and comprehensive. Clinics should also undertake contact tracing as soon as possible in the event of non-attendance, consider facilitating access to ART clinics and take loss to follow-up into consideration when assessing death rates.

Introduction

Many resource-poor countries with HIV epidemics are scaling up antiretroviral therapy (ART). By December 2005, an estimated 1.3 million people from low- and middle-income countries had been placed on treatment; 810 000 of these were in sub-Saharan Africa.¹ Treatment outcomes reported from various clinics in sub-Saharan Africa, Haiti, Asia and South America have been good, comparable with those observed in countries with higher incomes.^2–7 Patient outcomes are usually categorized as patients alive and on treatment, stopped treatment, transferred to another facility, dead or “lost to follow-up”. Depending on the facility, patients are classified as lost to follow-up if they have missed two or more consecutive clinic appointments² or have not been seen for at least 2 or 3 months.^4–6 The proportion of patients lost to follow-up differs between clinics, ranging from 5% in programmes supported by Médecins Sans Frontières⁵ to 25% in western Kenya.⁴ In 18 ART-delivery programmes in Africa, Asia and South America, 15% of 4810 patients were lost to follow-up in the first year of ART, with the average being 12% in programmes with active follow-up and 19% in those with no active follow-up.⁷ To our knowledge, there are no published data from resource-poor countries on the true outcome status of these patients.

Malawi – a small resource-poor country in southern Africa – has been engaged in ART scale-up for nearly 3 years. By 31 March 2006, 46 702 patients had ever started receiving ART in 66 facilities in the public sector; of these, 33 891 (73%) were registered as alive, 5131 (11%) were dead, 4226 (9%) were lost to follow-up, 3140 (7%) had transferred to another facility and 314 (1%) had stopped therapy.⁸ We hypothesized that many patients lost to follow-up may have died, and therefore conducted a study in northern Malawi to investigate this problem.

Methods

Details on the delivery and monitoring of ART in Malawi have been described previously.⁹ When patients start ART, their details are entered on master cards and an ART register. Patients attend the ART clinic each month, at which time their outcome status is entered on the master card and they are given another month’s supply of ART drugs. If a patient is not seen in the clinic for three consecutive months, the patient is registered as a “defaulter” (an abbreviated term for “lost to follow-up”) on the master card and also on the register.

Four public-sector ART facilities in the northern region of Malawi were selected for the study. These included one central hospital and three peripheral hospitals (two district hospitals and one mission hospital). Provision of free ART was initiated at the central and mission hospitals in July 2004, and at the two district hospitals in January and June 2005 respectively. For the period between the date when each facility started to provide free ART and 31 March 2006, we identified all patients indicated on the master cards and registers as a defaulter. Using contact details, ART facility staff visited the patients’ homes to try to ascertain their true outcome status. If patients were traced to the home, they were asked whether they were still taking ART and if not, why they had stopped. If the patient had died, the relatives were asked when the patient had died. If the patient had moved away, relatives or friends were asked when they had moved.

Ethical approval for the study was granted by the Malawi National Health Sciences Research Committee. Informed consent was obtained from patients being interviewed, and staff took care not to disclose that the patient was receiving ART when interviewing relatives or friends in the event of the patient’s death or transfer.

Data were collected on structured forms, and analysis was carried out using Epi Info version 6.0. Patients from the central and the peripheral hospitals were compared: the χ² test with relative risks (RR) and 95% confidence intervals (CI) were used for characteristics and outcomes, and the student’s t-test was used for time periods between start of ART, default and patient tracing, with differences at the level of 0.05 being regarded as significant.

Results

During the period of observation, 5009 patients were registered as ever having started ART in the four hospitals. Of these, 253 (5%) were registered as defaulters. Table 1 shows patient characteristics, reasons for starting ART, treatment regimens, results of defaulter tracing and time periods between start of ART, date of default and date of patient tracing.

Table 1. Characteristics and true outcome status of patients on ART who were classified as “lost to follow-up” from four hospitals in Malawi.

Characteristics and outcome status	All hospitals	Central hospital	Three peripheral hospitals
Number of patients ever started on ART	5009	3158	1851
Female patients	2985 (60%)	1854 (59%)	1131 (61%)
Children (aged < 15 years)	403 (8%)	371 (12%)	32 (2%)
Number of patients who defaulted	253 (5%)	126 (4%)	127 (7%)
Median (range) time between starting ART and diagnosis of default in months	4.3 (2.6–24.6)	3.7 (2.9–19.5)	4.5 (2.6–24.6)
Median (range) time between diagnosis of default and home visits in months	6.4 (0.1–18.3)	8.2 (0.1–18.3)	5.1 (0.1–15.6)
Characteristics of defaulting patients:
Median age (years)	35	35	35
Female patients	151 (60%)	67 (53%)	84 (66%)
Children (aged < 15 years)	6 (2%)	4 (3%)	2 (2%)
Indication for ART:
WHO clinical stage 3	172 (68%)	87 (69%)	85 (67%)
WHO clinical stage 4	72 (28%)	31 (25%)	41 (32%)
WHO clinical stage 1 or 2 with a CD4 count < 250 per mm³	9 (4%)	8 (6%)	1 (1%)
ART regimen:
First-line regimena	250 (99%)	124 (98%)	126 (99%)
Alternative first-line regimenb	3 (1%)	2 (2%)	1 (1%)
True outcome status at the default tracing visit:
Total alive	58 (23%)	18 (14%)	40 (32%)
Status of patients found to be alive:
Alive on ART at the same clinicc	1 (2%)	0 (0%)	1 (3%)
Alive on ART at a different clinicd	20 (34%)	9 (50%)	11 (28%)
Alive but had stopped ART	37 (64%)	9 (50%)	28 (69%)
Dead	127 (50%)	58 (46%)	69 (54%)
Time of death after last ART attendance:
Month 1	30 (24%)	15 (26%)	15 (22%)
Month 2	29 (23%)	15 (26%)	14 (20%)
Month 3	14 (11%)	5 (9%)	9 (13%)
Month 4 and beyond	54 (42%)	23 (39%)	31 (45%)
Lost to follow-up	68 (27%)	50 (40%)	18 (14%)
Reasons for unsuccessful tracing:
Incorrect address in ART register	55 (81%)	42 (84%)	13 (72%)
Patient moved and family remained	7 (10%)	4 (8%)	3 (17%)
Patient and family moved	6 (9%)	4 (8%)	2 (11%)

ART, antiretroviral therapy.
^a Fixed drug combination of stavudine, lamivudine and nevirapine.
^b Drug regimen where zidovudine is substituted for stavudine in case of peripheral neuropathy, or efavirenz for nevirapine in case of hepatitis or severe cutaneous reaction.
^c Patient came late for appointment and did not disclose to ART clinic staff that he/she was already registered for treatment.
^d Patients transferred to other ART clinics without informing the staff from the clinic where they had first been registered.

For all defaulters, the median time between start of ART and date of default was 4.3 months, and between date of default and home visits was 6.4 months. Overall, the commonest reason for default was death, this being the case for 50% of patients. Of those who died, 57% died within 3 months of their last clinic attendance, the median time between start of ART and death being 2.9 (range, 0.1–19.2) months. Fifty-eight (23%) patients were still alive; about one-third had transferred themselves to another ART clinic without informing the clinic where they were registered and about two-thirds had stopped therapy. Of the patients who had stopped ART, reasons included the high cost of transport to the clinic (13 patients; 35%), religious beliefs (4 patients; 11%), persuasion by relatives to stop ART (4 patients; 11%) and other reasons (16 patients; 43%). The remaining 68 patients (27%) could not be traced, most frequently because of an incorrect address in the ART register. The median time between default and home visits was 1.8 (range, 0.1–16) months for patients who were found to be still alive and 4.8 (range, 0.1–18.3) months for patients who could not be traced, a highly significant difference (P < 0.001).

The characteristics of patients from the central and peripheral hospitals were similar, except that there were significantly more female defaulters in peripheral than in central hospitals (P < 0.05). The time between default and home visits was significantly longer for patients from central hospitals than for patients from peripheral hospitals (P < 0.001). Significantly fewer patients from central hospitals were found alive (RR, 0.45; 95% CI, 0.28–0.75; P < 0.01) and significantly more patients from central hospitals could not be traced (RR, 2.80; 95% CI, 1.7–4.1; P < 0.001) compared with patients from peripheral hospitals. There were no significant differences in the reasons for stopping ART or for continued loss to follow-up between the two groups.

Discussion

This study shows that half of the patients receiving ART who were subsequently lost to follow-up were dead, with a large proportion dying soon after they failed to attend the clinic. The reasons for death were not ascertained. About one-quarter of patients were alive, some having transferred to another facility and some deciding to stop therapy. An important reason for stopping therapy was the cost of transport from homes to clinics. The remaining patients could not be traced, with an incorrect address in the register being the most common reason. Whether patients gave a wrong address or whether the details obtained by clinic staff were insufficient is unclear. These overall outcomes are similar in many respects to those found several years ago in Malawi in a study in patients registered as defaulters with regard to treatment for tuberculosis.¹⁰

There were some differences between the central and peripheral hospitals in terms of the characteristics of defaulters. A higher proportion of women from the peripheral hospitals were defaulters. Although part of the explanation may be that peripheral hospitals registered slightly more women to receive ART, other factors may be responsible and this requires further study. The rate of failure to trace patients was higher for patients from the central hospital, and this may relate to the longer period of time between default and the home visit and the general difficulties faced in tracing mobile patients in a congested urban environment.

There are several important lessons to be learnt from this study (Box 1). First, ART clinics need to ensure that the address recorded for the patient is correct in order to facilitate contact tracing if this becomes necessary. Second, provided that resources are available, clinics should attempt to trace patients who stop attending the clinic in order to bring patients back to therapy. This needs to be done as soon as possible, as the longer the delay the more likely it is that the patient will be untraceable. Third, ART programmes need to consider the plight of those who have difficulty accessing ART clinics, and either try to site new ART clinics in better locations, set up outreach clinics or look into ways of financially supporting patients’ transport to clinics. Fourth, there may be differences in outcomes between patients who have defaulted from central urban hospitals compared with those from more peripheral rural hospitals, and ART programmes should determine local causes of loss to follow-up. Finally, country and clinic reports on treatment outcomes must take the issue of loss to follow-up into consideration, particularly with regard to death rates. ■

Box 1. Lessons learned.

• An operational research study in northern Malawi found that the reasons for being “lost to follow-up” for patients started on antiretroviral therapy (ART) were: death, 50%; alive and on ART, mainly at another clinic, 8%; alive and stopped therapy, 15%; and unable to be traced, 27%. The commonest reason for remaining patients being “lost to follow-up” was an incorrect address in the ART register.

• ART clinics need to record correct and comprehensive addresses, and should try to do contact tracing as soon as possible in the event of clinic non-attendance. Clinics also should consider facilitating better access to care for those with transport difficulties.

• True outcome status may differ depending on whether the ART clinic is in an urban or rural environment.

• In any comparison of treatment outcomes between clinics, particularly with regard to death rates, the issue of patient loss to follow-up cannot be ignored.