Spastic diplegia in children with HIV encephalopathy: first description of gait and physical status

Authors

  • Nelleke G Langerak,

    Corresponding author
    1. Neurosurgery Division, Department of Surgery, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa
    2. Physiotherapy Division, Department of Interdisciplinary Health Sciences, Faculty of Medicine and Health Sciences, Stellenbosch University, Tygerberg, South Africa
    • Correspondence to Nelleke G Langerak, Division of Neurosurgery, H-53 Old Main Building, Groote Schuur Hospital, Observatory 7925, Cape Town, South Africa. E-mail: nellekelangerak@hotmail.com

    Search for more papers by this author
  • Jacques du Toit,

    1. Orthopaedic Surgery Division, Department of Surgery Sciences, Faculty of Medicine and Health Sciences, Stellenbosch University, Tygerberg, South Africa
    Search for more papers by this author
  • Marlette Burger,

    1. Physiotherapy Division, Department of Interdisciplinary Health Sciences, Faculty of Medicine and Health Sciences, Stellenbosch University, Tygerberg, South Africa
    Search for more papers by this author
  • Mark F Cotton,

    1. Children's Infectious Diseases Clinical Research Unit, Department of Paediatrics and Child Health, Stellenbosch University and Tygerberg Children's Hospital, Tygerberg, South Africa
    Search for more papers by this author
  • Priscilla E Springer,

    1. Neurodevelopmental Division, Department of Paediatrics and Child Health, Stellenbosch University and Tygerberg Children's Hospital, Tygerberg, South Africa
    Search for more papers by this author
  • Barbara Laughton

    1. Children's Infectious Diseases Clinical Research Unit, Department of Paediatrics and Child Health, Stellenbosch University and Tygerberg Children's Hospital, Tygerberg, South Africa
    Search for more papers by this author

Abstract

Aim

The aim of this study was to explore the physical status and gait patterns of children with spastic diplegia secondary to human immunodeficiency virus encephalopathy (HIVE).

Method

A cross-sectional study was conducted on children diagnosed with HIVE and spastic diplegia. Sociodemographic and clinical background information was obtained, followed by three-dimensional gait analysis (3DGA) and a physical examination including assessments of muscle tone, strength, motor control, contractures, and bony deformities of the lower extremities.

Results

Fourteen children (eight males, six females; mean age 5y 8mo [SD 9mo], range 4y 4mo–6y 10mo) were studied. The cohort was divided into two groups based on distinctive gait patterns. Nine participants in group I showed only limited abnormalities. Group II displayed a more pathological gait pattern including stiff knee and equinus ankle abnormalities. Results of 3DGA, as with the physical examination outcomes, showed increased impairments from proximal to distal (except for hip extension).

Interpretation

This study provides a first description of distinctive gait patterns and related physical characteristics of children with HIVE and spastic diplegia. Further research is necessary.

Abbreviations
3DGA

Three-dimensional gait analysis

ART

Antiretroviral therapy

BMI

Body mass index

FMS

Functional Mobility Scale

HIVE

Human immunodeficiency virus encephalopathy

PROM

Passive range of motion

SES

Socio-economic status

What this paper adds

  • Children with spastic diplegia due to HIVE present with physical impairments that are different to spastic diplegia related to CP.
  • There may be more than one neuropathogenic pathway in HIVE.
  • Further investigation of aetiology and natural history of spastic diplegia in HIVE is required.

Worldwide, approximately 34 million people are infected with human immunodeficiency virus (HIV). This is the highest number ever reported, and is most probably as a result of increased survival because of greater access to combination antiretroviral therapy (ART).[1]

Children acquire HIV primarily through mother-to-child-transmission. The most common clinical feature of HIV-related central nervous system disease is HIV encephalopathy (HIVE), which can be a presenting feature.[2] Primary and persistent infection of the brain manifests as progressive HIVE presenting with more severe neurological deficits. A less severe static (plateau) form occurs mainly in children on ART.[3, 4]

Early administration of ART dramatically reduces the risk of HIVE in children.[5, 6] Nevertheless, HIVE still produces significant morbidity among children living with HIV.[7] One of the residual impairments of HIVE is spastic diplegia.[8]

Clinicians have noted that the clinical manifestations of spastic diplegia secondary to HIVE are similar to cerebral palsy (CP). It is controversial as to whether these could be described as CP because the lesion in the brain may not be static, a requirement for diagnosing CP. Corticospinal tract signs due to HIVE can diminish on ART but residual spastic diplegia may persist.[8]

Motor deficits of abnormal muscle tone, motor control, impaired strength, and balance are described in CP. These abnormalities can lead to secondary bone deformities and impaired muscle and joint contractures, which often result in typical pathological spastic diplegic CP gait patterns (e.g. jump, crouch gait),[9] or other gait abnormalities.[10]

Three-dimensional gait analyses (3DGA) in combination with physical examination is an accepted objective measure to interpret motor deficits in ambulant patients with CP.[11] There is a paucity of literature on physical outcomes in HIV-infected children on ART. To our knowledge, the combination of a detailed medical history, physical examination, and the assessment of pathological gait patterns using 3DGA in children with HIVE-related spastic diplegia has not been described.

The aim of this study was to explore the physical status and gait patterns of children with spastic diplegia secondary to HIVE using 3DGA and a standardized physical assessment. This information could improve our understanding of the physical impairments of children with HIVE, providing objective information to orthopaedic and neurodevelopmental specialists for treatment options.

Method

Participants

This was a cross-sectional study of HIV-infected children attending an infectious diseases clinic or research unit at Tygerberg Children's Hospital, Cape Town, South Africa. Children with spastic diplegia were screened and referred by their clinicians for possible participation. A developmental paediatrician (BL) selected participants based on strict criteria after examining the children and reviewing their medical records.

Inclusion criteria were diagnosis of HIVE according to Centre for Disease Control criteria,[12] infection by mother-to-child-transmission, on ART, ambulatory with spastic diplegia, aged between 4 years and 10 years, currently living in the greater Cape Town area, and with a cognitive level sufficient to cooperate during assessments. Exclusion criteria were other neuromuscular or CNS disorders or insults including secondary causes of HIV-related CNS problems, birthweight less than 2000g, dystonic/athetoid/ataxia/hypotonic symptoms, previous orthopaedic and/or neurosurgical interventions, spasticity medication in the past 6 months, and previous botulinum toxin injections in the lower extremities.

Informed consent was obtained from the parents or legal guardians in their home language. In addition, assent was obtained from children older than 7 years and able to understand the process. The Health Research Ethics Committee of Stellenbosch University, South Africa, approved the study (ethics number N11/07/204), and the principles outlined by the Declaration of Helsinki were adopted in this study.

Outcome measures

Participants' sociodemographic and clinical background information was obtained during a structured interview, review of medical records, and standardized assessments. This information included sex, age, socio-economic status (SES), body mass index (BMI), previous physiotherapy treatment, medical history, and details of ART. SES was calculated as the number of people living in the house divided by the number of rooms (excluding the kitchen and bathroom) in the house.[13] Low SES was defined as a housing ratio of greater than 1.5 (e.g. five people living in a three-roomed house), and high SES as a ratio less than 1.0. BMI was calculated from the participant's bodyweight and height, measured on the day of assessment.

Age when participants started walking was recorded. Current status was defined using the Gross Motor Function Classification System (GMFCS) from levels I to V, describing increasing functional mobility limitations, with level V the worst.[14] In addition, walking ability was quantified using the Functional Mobility Scale (FMS), an outcome measure that assesses the child's ability to achieve daily activities and interact with society.[15] This scale consists of a six-level ordinal grading system, ranging from 1 (use of wheelchair) to 6 (independent on all surfaces). Participants were rated for three distances: 5m, 50m, and 500m (e.g. in house, school, and outside). Distances too far for a child to walk were recorded as ‘unable’.

Based on information from participants' medical records, age at diagnoses (HIV and HIVE) and ART history (age started, time of interruption, adherence, and type used currently) were extracted. Virological control was categorized into good (undetectable viral loads), moderate (slightly raised viral loads, mostly suppressed), or poor (virological failure), according to laboratory reports or clinical notes.

Computerized 3DGA was performed using an eight-camera Vicon optoelectronic system (Oxford Metrics, Oxford, UK) synchronized with frontal and sagittal digital video cameras. Sixteen reflective markers were used as described for the lower body plug-in-gait marker placement and model. Participants wore shorts and walked barefoot at their own self-selected speed over a 15m walkway. At least five trials were captured and processed, of which three trials of good quality were selected for further analysis. Outcomes measured included pelvic, hip, knee, and ankle joint sagittal plane kinematic graphs presented per stance and swing phase of the gait cycle.

A paediatric orthopaedic surgeon (JDT) and two physiotherapists (NGL, MB) conducted the history taking and physical assessment. The measuring technique and terminology used were according to the manual of the American Academy of Orthopaedic Surgeons.[16] Parameters included (1) muscle tone; (2) muscle strength; (3) selective motor control; and (4) contractures, lever arm dysfunction, and other bony deformities of lower extremities.

The degree of muscle tone was indicated using the Ashworth Scale, ranging from score 0 (no increase in tone) to 4 (limb rigid in flexion and extension).[17] Muscle strength was assessed using the Medical Research Council scale ranging from 0 (no strength) to 5 (normal).[18] Selective motor control was assessed using a selectivity scale ranging from 0 (no ability) to 2 (complete ability to isolate movement).[19] In the case of a fixed contracture, which limited the assessment, the outcome of the physical examination was indicated as ‘contracture’.

Contractures, lever arm dysfunction, and other bony deformities were assessed by hip, knee, and ankle passive range of motion (PROM) measurements performed with a goniometer, relative to standardized anatomical landmarks. For knee extension PROM measurements, the standard assessment with extended hips was performed and the popliteal angle shift was calculated. The standard knee extension PROM measurement indicates true knee capsular tightness, whereas the popliteal angle shift provides information about excessive anterior tilt due to tight hip flexors, weak hip extensors, and/or weak abdominal musculature. This shift, also called the hamstring shift, is calculated by the difference between the unilateral popliteal angle (functional hamstring contracture) and the bilateral angle (true hamstring contracture).[20, 21] The assessment for lever arm dysfunction included measures of femoral anteversion (Craig's test),[22] and tibial torsion (bi-malleolar axis and thigh-foot angles assessments).[23] Other bony abnormalities indicated were patella alta and foot deformities.

Data analysis

Participants were divided into groups based on distinctive gait patterns quantified by observational screening of frontal and sagittal videos and 3DGA outcomes. Special indicators used during the video analysis were posture, motor control, and lower extremities joint range of motion. 3DGA gait patterns were observed with attention to shape and magnitude of the graphs and timing of events within the gait cycle. This grouping procedure was performed in consensus with the experienced gait analysts (NGL, JDT, DF, MB), who were blinded to participants' background information including sociodemographic and clinical characteristics.

Background information was described for each participant stratified per group. Outcomes of the physical examination and 3DGA were described per group. Frequencies were used for categorical variables and medians with interquartile ranges for continuous variables (all non-Gaussian).

Results

Participants

Fourteen children (eight males, six females; mean age 5y 8mo [SD 9mo], range 4y 4mo–6y 10mo) fulfilled the selection criteria and participated in this study. Half of the group was classified as GMFCS level I, with six participants as level II, and one as level III. This was also reflected in the FMS scores, with most participants scoring 6 (independent on all surfaces) or 5 (independent on level surfaces) for 5m, 50m, and 500m distances. All children had BMIs within the normal range. Eight participants had a background of low SES, five were middle, and only one classified as high SES (Table 1).

For the clinical characteristics, no children were born prematurely. Five children had previously been treated for pulmonary tuberculosis without sequelae. No epilepsy was documented. Two children previously received oral baclofen for spasticity, but this was discontinued due to a lack of treatment response. No other medication related to the neuro-musculo-skeletal system was noted. Most parents or primary caregivers reported that their child had received some form of physiotherapy in the past (n=12) and/or recently (n=3). However, this was neither as frequent nor intense as required, because of financial constraints.

The gait analysts identified two typical gait patterns, with nine participants clustered in group I, and five combined in group II. Sociodemographic and clinical features were equally distributed for the two groups (Table 1). The age at diagnosis of HIV ranged from 4 weeks to 11 months, and diagnosis of HIVE from 5 months to 2 years 10 months, although data were missing for two children. Apart from one child, all participants started ART at 12 months of age or younger (range 2–24mo), with group I starting at a lower mean age than group II (6.7mo and 11.2mo respectively). At the time of assessment, the most common ART was lamivudine (3TC) and ritonavir-boosted lopinavir (LPV/r) combined with either zidovudine (AZT) or stavudine (d4t). One child was on 3TC monotherapy, owing to poor compliance and persistent viral failure. ART had been interrupted in four children (group I: numbers 2, 3, 4, and 6) for a period of 2 weeks to 7 months under strict supervision in a research study.[24, 25] Viral loads and CD4 counts were not recorded routinely in all children. However, there were three children with problems of adherence and treatment failure (one child of group I, two of group II), whereas viral suppression was documented in eight children (five children of group I, three of group II). Group size was too small for meaningful comparison.

Table 1. Sociodemographic, clinical, and walking characteristics of study cohort
NumberSociodemographicsWalking abilityDiagnosisART history
SexAge (y:mo)SES equationBMIWalking onset age (y:mo)GMFCS levelFMSHIV ageHIVE age (y:mo)Started age (mo)AdherenceType at assessment
5m50m500m
  1. ART, antiretroviral therapy; SES, socio-economic status; BMI, body mass index; GMFCS, Gross Motor Function Classification System; FMS, Functional Mobility Scale; HIV, human immunodeficiency virus; HIVE, human immunodeficiency virus encephalopathy; AZT, zidovudine; 3TC, lamivudine; LPV/r, lopinavir/ritonavir; d4t, stavudine; NVP, nevirapine; DDI, didanosine; ABC, abacavir; N/A, no data available.

Group I
1Male6:14.015.01:0I6668wks2:89GoodAZT/3TC/LPV/r
2Female5:102.016.11:6I6665wks2:73GoodAZT/3TC/ LPV/r
3Female5:51.314.52:0I66Unable5wks1:72GoodAZT/3TC/ LPV/r
4Male6:12.015.43:0I6662mo1:34Goodd4t/3TC/LPV/r
5Male5:41.016.43:0II5552mo0:99N/AAZT/3TC/ LPV/r
6Male4:42.014.62:0II5554mo1:24Moderated4T/3TC/LPV/r
7Female6:53.014.65:0III54Unable11mo0:1112Moderated4T/ 3TC/LPV/r
8Male6:03.514.51:6II55Unable7wks1:09GoodAZT/3TC/NVP
9Female5:91.015.82:0I6664wks1:48PoorAZT/3CT/ LPV/r
Mean5:82.215.2
Group II
10Female5:91.815.03:0I6666wks0:66GoodAZT/3TC/ LPV/r
11Male6:51.515.34:0II55UnableN/A2:109GoodAZT/3TC/ LPV/r
12Male4:60.716.33:0II5554mo0:55Poor3TC
13Female6:101.216.43:0II55510mo1:512PoorAZT/3TC/DDI/LPV/r
14Male4:73.017.21:6I666N/Ax24GoodABC/3TC/LPV/r
Mean5:71.616.0

Gait analysis

Figure 1 shows the sagittal gait graphs of groups I and II and a normal gait pattern. Participants in group I walked with a gait pattern close to the pattern of typical developed persons.[26] On average, the pelvic tilt, hip, and ankle plantar- and dorsal flexion fell within the normal ranges. The knee pattern was also similar to normal, but showed on average slightly decreased and delayed flexion during the swing phase.

Figure 1.

Sagittal gait patterns. Light blue band represents mean (1SD) for typical developed persons,[26] whereas the blue lines show the mean (1SD) for group I and the red lines for group II.

Group II showed a more pathological gait pattern, particularly for the knee and ankle. The knee maintained a flexed position and decreased range of movement during the entire gait cycle with a delayed peak swing knee flexion in mid- to terminal swing. The ankle showed a continuous plantar flexion ankle pattern during both phases of gait. Pelvic and hip pattern for group II was more comparable to typical developed persons and group I, although less stability in pelvic tilt was noted.

Physical examination

Lower limb muscle tone was higher in group II than in group I (Fig. 2). Group I had mainly an increased plantar flexor tone response, whereas group II had a progressive increase in tone from proximal to distal. Three children from group II had contractures around the knee and/or ankle, suggesting prolonged periods of increased and abnormal tone. Both groups showed a decrease in strength and selectivity in a proximal to distal distribution (except for hip extension movement), with group II being much more severely affected (Fig. 3).

Figure 2.

Muscle tone. Ashworth scale: 0, no increase in tone; 1, slight increase in tone giving a catch when the limb is moved in flexion or extension; 2, more marked increase in tone but limb easily flexed; 3, considerable increase in tone, passive movement difficult; 4, limb rigid in flexion and extension.[17]

Figure 3.

Muscle strength and selectivity. (a), group I strength. (b), group II strength. (c), group I selectivity. (d), group II selectivity. Strength Medical Research Council scale: 2, motion with gravity eliminated; 3, antigravity strength; 4, some resistance present; and 5, normal strength. Selectivity scale: movement observed: 0, only patterned; 1, partly isolated: 2, completely isolated.[18] Eversion (long), eversion by m Peroneus Longus; Eversion (brev), eversion by m Peroneus Brevis.

PROM for hip flexion was normal for both groups (Table 2). However, there were limitations in hip extension, but only one participant in group II had a severe flexion contracture of greater than 30°. Functional hip abduction range (>15° with hip in extension) was present in both groups. The Duncan-Ely static test for tight rectus femoris muscle was negative in both groups. For the PROM of the knee, group I showed no limited knee extension, whereas two participants from group II had capsular contractures. These two children had functional hamstring contractures (unilateral popliteal angle) of greater than 30° and displayed a hamstring shift of less than 20° (bilateral popliteal angle). PROM of the ankle in plantar flexion was normal in both groups, whereas group II had significant ankle equinus with predominant influence from the gastrocnemius muscle (positive Silfverskoïld Test).

Table 2. Passive range of motion assessments for groups I and II
 Group I (n=9)Group II (n=5)
Median (first and third interquartile ranges)Median (first and third interquartile ranges)
Hip
Flexion140 (133–150)133 (130–139)
Extension (Thomas test)0 (−5 to 0)0 (−19 to 0)
Abduction40 (30–40)25 (20–30)
Adduction30 (20–30)20 (20–22)
External rotation45 (40–50)30 (30–46)
Internal rotation50 (40–50)40 (40–48)
Femoral anteversion25 (20–30)20 (15–25)
Knee
Extension0 (0)0 (−14 to 0)
Popliteal angle:
Unilateral5 (0–10)18 (5–38)
Bilateral0 (0–5)8 (0–34)
Shift3 (0–5)5 (1–10)
Bimalleolar axis20 (16–22)15 (−15 to 20)
Thigh–foot angle17 (10–20)9 (0–10)
Ankle
Dorsiflexion with:
knee extension10 (6–15)−11 (−18 to −1)
knee flexion20 (15–29)4 (−7 to 18)
Plantar flexion60 (50–60)60 (60–68)

None of the participants from group I displayed rotational or bony deformities, whereas three children in group II showed rotational malalignment (Table 2). One participant had severe bilateral femoral anteversion (>30°) as well as an increased internal hip rotation, and two participants had increased internal tibial torsion. As expected for participant's age and weight, minimal foot deformities and no patella alta were present.

Follow-up management

As a result of the physical examination and 3DGA, the paediatric orthopaedic surgeon (JDT) decided that no participants from group I required orthopaedic intervention. Two participants from group II were referred for botulinum toxin injections in their medial hamstrings and gastrocnemius muscles, whereas the remaining three were referred for orthopaedic review to consider future surgery intervention.

Discussion

This study presents the first gait and physical characteristics of children with HIVE and spastic diplegia on ART. Gait analysis experts were able to divide the study cohort into two groups. Group I (64%), walked with a gait pattern close to normal,[26] whereas group II had a more pathological gait pattern. These results of 3DGA were in line with the outcomes of the physical examinations reflecting participants' physical status for both groups and supports differential expression of HIVE.

The gait patterns of groups I and II could not be classified in the typical spastic diplegic CP gait patterns described by Rodda et al.[9] (Table 3). Group I showed only limited abnormalities, and group II was outside the ranges for each level described for typical jump, crouch, true or apparent equinus gait. The position of the pelvis was close to normal (some anterior tilt)[26] and the two mild bumps seen in the gait graphs are common in each of the typical CP gait patterns.[9] The hip graphs also showed ranges very close to normal, including hip extension during late stance as described for true equinus gait. On the other hand, the excessive knee flexion throughout the gait cycle is recognized in jump and apparent equinus and crouch gait, while the ankle plantar flexion is typical in true equinus and jump gait.[9]

Table 3. Typical gait pattern of children with cerebral palsy and spastic diplegia
 True equinusJump gaitApparent equinusCrouch gait
  1. Adapted from Rodda et al.[9] aAlso indicated in group II.

  2. a

    Also indicated in group II.

PelvisAnterior tilted; normal rangeaAnterior tilted; normal rangeaAnterior tilted; normal rangeaAnterior tilted; normal range; posterior tilteda
HipNormal extensionaFlexedFlexedFlexed
KneeNormal extension; recurvatumFlexedaFlexedaFlexeda
AnkleEquinusaEquinusaNormal rangeDorsiflexed

Wren et al. described gait abnormalities in 291 children with diplegia due to CP. When these definitions were applied, group I did not show typical abnormalities. Children in group II walked with a ‘stiff knee’ defined as follows: decreased knee range of motion from knee extension in stance to knee flexion in swing phase and/or delay in maximum knee flexion mid- to late swing phase. In addition, equinus ankle was present, recognized by an increased ankle plantar flexion (with or without varus/valgus foot) during stance phase. These gait abnormalities were in line with the physical examinations in group II. The stiff knee gait abnormality can be due to increased muscle tone in knee flexors (hamstrings), impaired strength and motor control in knee extensors (rectus femoris dysfunction), functional hamstring tightness, and/or capsular knee flexion contractures. The equinus ankle could be due to increased muscle tone in plantar flexors and/or ankle contractures.

The physical status of children, particularly from group II, showed increased impairments from proximal to distal except for the hip extension movement. This could suggest a myotomal distribution: hip flexors (L1–2); knee extensors (L2–3); ankle dorsiflexors (L4); long toe extensors (L5); and hip extensors, ankle plantar flexors, and foot everters (S1).[27] However, the study cohort was too small and diverse to attempt mimicking this myotomal distribution. Considering the low frequency of physiotherapy sessions the children attended, clinical signs may appear to be progressive as spasticity and increasing joint deformity may be due to the mechanisms associated with static lesions and not HIVE.

We were unable to identify clinical correlates between the two groups to explain differences in gait patterns. We were also unable to separate the participants according to severity, or assign them to three well-described patterns of HIVE summarized by Civitello:[2] (1) subacute progressive encephalopathy, often seen when not yet on ART (can be spastic quadriplegia or diplegia); (2) plateau type progressive encephalopathy, more indolent with either the absence of, or a slower rate of acquiring new skills (motor involvement, especially spastic diplegia, is common); and (3) static encephalopathy where children have fixed neurodevelopmental deficits including motor dysfunction. The neuropathological pathways underlying the different clinical manifestations in each child are due to a unique combination of risk factors including maternal HIV viral load and ART during pregnancy, timing of infection, underlying immune response, and timing of ART.28

Our study is limited because of its cross-sectional design with retrospective record review for participants' background information. The timing of HIV infection (pre- or perinatal), a more detailed description of virological or immunological parameters, their response to ART, and/or neuroimaging may have provided an understanding of the differences observed in physical and gait characteristics between the groups. It is possible that group II represents a more severe form of encephalopathy, but we were unable to assess this.

In conclusion, this is the first study describing the gait patterns and physical status of children with HIVE and spastic diplegia on ART. Two distinct groups were observed: one had limited impairments, whereas the other group walked with a stiff knee and equinus ankle. The limitations seen in the gait patterns were confirmed on the physical assessments, including increased number of impairments for the second (more impaired) group. The gait patterns were different compared with typical spastic diplegia patterns of CP with static encephalopathy, which should be considered when treating the secondary impairments from encephalopathy.

This study provides a clear documentation of physical impairments in children with HIVE and spastic diplegia on ART, which we feel is under-reported. We therefore encourage longitudinal clinical, gait, and neuroimaging studies to establish a better understanding of the underlying neuropathological mechanisms and natural history of HIVE and spastic diplegia. This research should bring HIVE to the attention of orthopaedic and neurodevelopmental specialists, and result in evidence-based guidance for optimizing management of physical impairments in children with HIVE and spastic diplegia.

Acknowledgements

We acknowledge the superintendent of Tygerberg Hospital for permission to conduct the study. H Rabie and S Innes made substantial contributions to the concept, design, acquisition, and revision of the paper for intellectual content. D Fisher contributed to data collection and interpretation of data. L Khethelo contributed as a translator during recruitment and assessments. We also thank the Harry Crossley Foundation for providing financial support to our operating expenses for this study.

Conflict of interests

The authors have stated that they had no interests that might be perceived as posing a conflict or bias.

Ancillary