Economic evaluation of a task-shifting intervention for common mental disorders in India
Christine Buttorff a, Rebecca S Hock b, Helen A Weiss c, Smita Naik d, Ricardo Araya e, Betty R Kirkwood f, Daniel Chisholm g & Vikram Patel f
a. Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, United States of America (USA).
b. Department of Mental Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, USA.
c. MRC Tropical Epidemiology Group, London School of Hygiene and Tropical Medicine, London, England.
d. Sangath Centre, Porvorim, India.
e. Academic Unit of Psychiatry, University of Bristol, Bristol, England.
f. Centre for Global Mental Health, London School of Hygiene and Tropical Medicine, Keppel Street, London, WC1E 7HT, England.
g. Department of Mental Health and Substance Abuse, World Health Organization, Geneva, Switzerland.
Correspondence to Vikram Patel (e-mail: email@example.com).
(Submitted: 24 February 2012 – Revised version received: 21 August 2012 – Accepted: 22 August 2012 – Published online: 14 September 2012.)
Bulletin of the World Health Organization 2012;90:813-821. doi: 10.2471/BLT.12.104133
Non-communicable diseases account for a growing burden on the health systems of developing countries. The effective management of these diseases typically requires a collaborative effort across the health workforce as well as continuing care for months or even years. In resource-poor areas, a “task-shifting” strategy can be beneficial, in which community or lay health workers (with oversight from primary-health-care practitioners and specialists) provide “front-line” care, instead of physicians and trained nurses.1 There is growing evidence of the effectiveness of such task-shifting in the management of some chronic conditions, including infection with the human immunodeficiency virus (HIV) or acquired immunodeficiency syndrome (AIDS),2,3 diabetes4 and some mental disorders.5–9 Although the strategy appears particularly attractive in the many low-income countries with inadequate numbers of physicians and trained nurses, there is considerable institutional resistance to the widespread implementation of the strategy and also concern that the quality of care will deteriorate.10 There is a clear need for more studies comparing the health outcomes of patients attended by lay health workers with those of patients attended by physicians and trained nurses. There is also a need for more studies in which the cost–effectiveness of the task-shifting strategy is evaluated.10
Depression and anxiety, two of the most prevalent non-communicable disorders, are often encountered in primary-care settings.11 Depression is predicted to become the leading cause of disability-adjusted life years by the year 2030.12 Depressive and anxiety disorders are classified separately in the tenth revision of the International statistical classification of diseases and related health problems (ICD-10).13 In public-health interventions, however, they are often grouped together as “common mental disorders” because they show a high degree of comorbidity, have similar epidemiological profiles and respond to similar treatments.11,14–16
In several studies, collaborative stepped care led by lay health workers has been found to be successful in the primary care of depression and/or anxiety in low- or middle-income countries.17–19 This approach encourages the most effective sharing of tasks between medical, specialist and non-medical staff. There are various “steps” or levels of treatment, with the most intensive treatments reserved for the most severe cases. Used together, the collaborative-care and stepped-care components of this strategy can maximize the efficient use of scarce resources, especially in those public health facilities where case management has previously been relatively poor.20 In the MANAS trial, the effectiveness of this approach in the primary care of patients with depression and/or anxiety was investigated in Goa, India. The design, implementation and general effectiveness of this cluster-randomized controlled trial have been described in detail elsewhere.18,19,21 Both public and private facilities were included in the trial because in India’s private facilities, the quality and costs of care are both generally higher than in public facilities. For example, private facilities offer repeated consultations with the same physician and are primarily financed with out-of-pocket payments from the patients. In contrast, many patients attending a public facility may see a different physician on each visit but will not pay for any of the consultations.
The present study evaluates the cost–effectiveness and cost–utility of the MANAS trial. We hoped that the additional resources needed to train, pay and supervise the lay health workers used in the “task-shifting” approach to the primary care of common mental disorders would promote recovery and reduced disability in a more cost–effective manner than more conventional care. In any particular country, the CHOICE programme of the World Health Organization (WHO) deems an intervention to be highly cost–effective if it generates an extra year of healthy life for an amount no greater than the country’s per capita gross domestic product.22
The present study formed part of the MANAS trial, which has been registered with Clinicaltrials.gov (NCT00446407) and previously described in detail.18,19,21 In brief, the trial was carried out in the state of Goa, in western India. Participants who met the initial eligibility criteria (e.g. aged > 17 years and spoke one of the four study languages) were screened for depression and/or anxiety by means of a pre-tested General Health Questionnaire.23 Subjects found positive for either of these common mental disorders were invited to participate. The trial comprised two consecutive phases: an evaluation of the task-shifting intervention in 12 public centres for primary health care that were operated by the government of Goa (Phase 1), and an evaluation of the same intervention in 12 private general practitioner clinics (Phase 2). In each phase, health-care facilities were randomized to the intervention arm (i.e. collaborative and stepped care) or the control arm (i.e. enhanced usual care, described later).
Subjects with depression and/or anxiety in the intervention arm received both collaborative care and stepped care. The collaborative care of each eligible subject was provided by three key health-care providers: the existing, full-time physician at the facility, a full-time lay health worker (or “health counsellor”) trained to provide psychosocial interventions, and a mental health specialist who visited each study facility once or twice a month. For stepped care, the intensity of the care offered to a subject was matched with the severity of that subject’s disorder, to optimize the use of scarce resources. Consenting subjects in the intervention arm were educated about their symptoms and the link between depression, anxiety and interpersonal difficulties. They were also taught strategies to reduce their symptoms (e.g. relaxation breathing exercises and scheduling activities) and provided with tailored information about the relevant social and welfare organizations. Subjects with mild depression and/or anxiety who did not respond well to such psycho-education were offered antidepressants and/or interpersonal therapy, as were subjects with moderate to severe depression and/or anxiety. The interpersonal therapy focused on the subject’s relationships with other people and their coping with events such as role transitions, conflict and grief. Case management, with proactive monitoring of outcomes and adherence support, formed the backbone of the intervention.
Eligible subjects in the control arm received “enhanced usual care”. For this, the existing, primary-care physician in the facility was provided with the results of the initial screening and a treatment manual. Physicians were permitted to administer the treatments of their choice but did not have access to any additional (i.e. trial-related) human resources.
The outcomes recorded for each subject 2, 6 and 12 months after recruitment were a psychiatric symptom score, presence/absence of either depression or anxiety and days of lost or reduced work. Quality-adjusted life years (QALYs) over the 12 months of follow-up were generated from disability scores.
For each subject, at enrolment and at each follow-up, a psychiatric symptom score between 0 and 49 was evaluated using the Revised Clinical Interview Schedule, which measures 14 symptom groups of common mental disorder, including depression and anxiety. Although this schedule was developed in the United Kingdom of Great Britain and Northern Ireland,24 it has been used in several studies in India.25–27 The PROSQY software package – a diagnostic algorithm based on the ICD-10 criteria for the diagnoses of common mental disorders28 – was used to determine whether a subject had a common mental disorder at each follow-up.
The 12-item short-form version of WHO’s Disability Assessment Schedule29 was used to assess disability outcomes. Standardized summary scores were converted to a preference-weighted utility index,30 which was then used to compute the additional number of QALYs generated by the intervention over the 12-month follow-up.
The answers to two questions in WHO’s Disability Assessment Schedule29 can be used to estimate the number of days in the previous month that someone was completely unable to work or able to work only part time because of a health condition. For the present study, these two numbers were summed to give the number of days in which working hours were reduced because of poor health. As decreases in psychiatric symptom scores or in the number of days of lost work would both be favourable outcomes, the reciprocals of these scores and numbers were used for the regression analyses, so that the resultant incremental cost–effectiveness ratios would be easier to interpret.
At each of the three follow-ups, subjects self-reported their health-care utilization, medication use and out-of-pocket payments on the same type of cost-of-illness inventory used in several earlier studies in India.25,31,32 Two key categories of costs were estimated: health-system costs (including those related to the intervention itself, comprising the costs of inpatient and outpatient care, medications and clinical investigations) and the “time costs” for the subjects and their families (i.e. the opportunity costs of time spent travelling to, waiting for or receiving care, plus the wages from any days of work lost).
The additional human resource use associated with the lay health workers employed in the intervention was evaluated using the clinical process indicator records listed in Appendix A (available at: http://sangath.com/images/file/MANASCEA_AppendixA.pdf). Some of the costs of the intervention were estimated by multiplying the total number of minutes a subject in the intervention arm had contact with a lay health worker by the per-minute cost of the health worker (Appendix A). All costs were calculated in Indian rupees (INR) for 2009 but are reported in United States dollars (exchange rate: INR 46.5 = US$ 1). As costs (and effects) for each subject were only followed for 1 year, no annual discounting was required.33
Costs, scores and days-of-work data from each follow-up were summed to give totals for the full year post-recruitment. Descriptive means and standard deviations (SDs) for the cost data are presented. To assess the differences between the treatment and control groups and to allow for the skewed nature of the data on costs and outcomes, generalized linear models were used, with gamma distributions and log transformations.34–36 Estimates were adjusted for baseline depression scores and are presented with confidence intervals (CIs). The data analyses used in earlier studies on the MANAS trial were based on cluster-level summaries37 but there were too few clusters to use this approach in the present study. Furthermore, the approach has been deemed inappropriate for analyses of cost–effectiveness.38 The analysis of resource use was based only on the data from subjects who were available for each of the scheduled follow-ups.39 In terms of subject age, gender and the other variables recorded, loss to follow-up appeared to be a random process. The substitution of missing data on costs with the corresponding mean, minimum or maximum values had no impact on our main conclusions (Appendix A). Cost–effectiveness acceptability curves,40 which show the probability that an intervention remains cost–effective at increasing monetary values,41 were plotted. All statistical analyses were performed using version 11 of the STATA software package (StataCorp. LP, College Station, United States of America).
Overall, 20 352 subjects were screened for depression and/or anxiety in the 24 study facilities. Of these, 3816 were found positive and 3434 met the criteria for eligibility; 2796 (81%) of the eligible subjects (1436 and 1360 subsequently assigned to the control and intervention arms, respectively) agreed to participate. In public facilities, 1437, 1416 and 1386 subjects were available at the first, second and last follow-ups. In private facilities, the corresponding numbers were 1054, 1013 and 981, respectively. Complete data were recorded for 1243 (75.4%) of the subjects recruited in public facilities and for 938 (81.7%) of those recruited in private facilities.
The descriptive means for costs are presented in Table 1. The mean total cost of the human resources associated with the intervention was INR 93 (US$ 2; SD: 1.53) per participant, or 2% of the unadjusted total health system cost incurred in the intervention arm. Subjects in the intervention arm had contact with the lay health workers, either in person or over the telephone, on a mean of 6.9 occasions (SD: 3.80) and for a mean total of 70.8 min (SD: 54.2).
Table 1. Average annual costs, per subject, in the intervention and control arms in public facilities, Goa, India, 2009
The regression-adjusted cost differences between the intervention and control arms are presented, for those with complete data, in Table 2. Total health system costs were marginally higher in the intervention arm than in the control arm but this difference was not statistically significant. However, overall time costs for subjects and their families and total costs were significantly lower in the intervention arm than in the control arm (P < 0.001 and < 0.01, respectively).
Table 2. Between-arm difference in annual costs, per subject in public facilities, and health outcomes, Goa, India, 2009
Over the 12 months of follow-up, mean psychiatric symptom scores improved by 3.84 points (95% CI: 3.29 to 4.38) more in the intervention arm than in the control arm (Table 2). Furthermore, compared with their counterparts in the control arm, subjects in the intervention arm gained significantly more QALYs and achieved significantly more days of work (Table 2).
Cost–effectiveness and cost–utility
Although negative incremental cost–effectiveness ratios can be difficult to interpret,36,42 the ratios calculated in the present study indicate that the intervention was both less costly and more effective than enhanced usual care in terms of all the health outcomes investigated. The between-arm difference in QALYs gained appeared small (0.02), partly because it only relates to a single year, but this difference represents a mean of 7.3 additional days free of depression and/or anxiety for each subject in the intervention arm. The mean health system cost per case recovered at the end of follow-up (Appendix A) was INR 5959 (US$ 128; 95% CI: 105 to 157) in the intervention arm and INR 6933 (US$ 149; 95% CI: 131 to 169) in the control arm. The between-arm difference in the total costs per case recovered was even more striking, with such costs in the public and private facilities about INR 5600 (US$ 120) and INR 4000 (US$ 86) lower, respectively, in the intervention arm than in the control arm.
Uncertainty analyses can provide a range of conditions over which an intervention is plausibly cost–effective.33,43 In the public facilities, incremental cost–effectiveness ratios indicate that the intervention would be cost-saving under about half of the conditions illustrated in Fig. 1. Although the intervention would be more effective under the other conditions, it would also be more costly.
Fig. 1. Total costs versus outcomes at 12 months, in public facilities, Goa, India, 2009
Uncertainty in the cost–effectiveness of an intervention can also be illustrated as a cost–effectiveness acceptability curve (Fig. 2).44 There are no accepted threshold values to indicate willingness-to-pay for improvement in depression measures such as the psychiatric symptom score, but the probability that the intervention will be cost–effective at very low threshold values of < INR 400 (US$ 8.60; i.e. less than the amount earned by an individual working for 3 days for the legal minimum wage in Goa45) is very high (nearly 1.00). When the only outcome considered is the number of days of work gained, the intervention always appears to be cost–effective. When time costs are included, the intervention appears to be cost-saving, since it improves health outcomes while lowering costs.
Fig. 2. A cost–effectiveness acceptability curve for health system costs at 12 months, in public facilities, Goa, India, 2009
The results of other sensitivity analyses indicated that the cost data were sensitive to the missing observations (Appendix A). However, the cost differences between the intervention and control arms were found to be consistently statistically significant, in favour of the intervention arm, under all conditions except the worst-case scenario.
In the private facilities that were studied, none of the between-arm differences seen in health outcomes reached statistical significance (Appendix A). Although, per subject, mean health system costs and total costs in the intervention arm were INR 916 (US$ 20) and INR 1511 (US$ 32) lower, respectively, than the corresponding values in the control arm over the year of follow-up, the associated 95% CIs (in INR, −3426 to 1110 and −4221 to 1008; in US$, −74 to 24 and −91 to 22, respectively) both crossed zero. This is why we present the findings for public facilities only in the main body of the paper.
Despite the additional resources required for the intervention led by lay health workers, the health system costs incurred over the 12 months of follow-up were similar across the two arms. In the public (but not the private) facilities investigated, time costs were lower and health outcomes were significantly better in the intervention arm than in the control arm. In the public primary-care facilities, therefore, the intervention appeared to be not only cost–effective but also cost-saving; the subjects in the intervention arm used and/or lost less cash and showed greater improvement in their mental state than the control subjects. There were no statistically significant between-arm differences in any of the health outcomes investigated in private facilities, probably because the standard of routine care in such facilities (i.e. the basic level of care experienced in the control arm) was relatively high. In these facilities, however, the care of the subjects with depression and/or anxiety was cheaper in the intervention arm than in the control arm and therefore the intervention still appeared advantageous from a cost-minimization perspective.
The use of task-shifting to reduce the barriers posed by shortages of mental health professionals is becoming increasingly common. One study has already shown it to be an effective approach.46 The present results indicate that such task-shifting can reduce the total costs of the care of patients with depression and/or anxiety and improve health outcomes in public facilities. In such facilities the intervention was cost–effective by WHO’s CHOICE programme criteria.22
Our study adds to the little that is known about the cost–effectiveness of task-shifting interventions for non-communicable or chronic diseases in developing countries. There is a clear need for more studies on this topic.6 A task-shifting pharmaceutical intervention for HIV/AIDS patients in South Africa was found to be cost–effective.47 There appears to have been only one previous investigation of the cost–effectiveness of a task-shifting intervention for the treatment of mental disorders in a developing country: in Chile, an intervention based on the stepped care of depression in women was found to increase health-system costs but provided an extra depression-free day for a small incremental cost of about US$ 1 per woman.17
The present study has several limitations. All the data on service utilization were participant-reported and therefore subject to recall bias. A recent review of self-reported service utilization in 42 studies identified several key factors that can influence the quality and accuracy of self-reported data, such as the sample population and cognitive ability, the recall time frame, the type of utilization, questionnaire design, the mode of data collection and the use of memory aides and probes.48 Most of these issues were addressed and mitigated in the present study, whose design was based on the Client Socio–Demographic and Service Receipt Inventory – European Version.49 A second limitation of the present study is that no baseline data on resource use were collected, although there were no significant between-arm differences in any baseline outcome measures. A third limitation is that detailed results are only presented for the public facilities, since the results from the private facilities did not show that the intervention was more effective. The full results from the private facilities are, however, available in Appendix A and other articles.18,19 Finally, our use of the minimum wage as a measure of the economic value of lost time was designed to be conservative but is subject to uncertainty. For example, there are some highly skilled workers in the study population who, presumably, earn much more than the minimum wage. There are also variable ways in which households cope with illness. A more detailed microeconomic analysis of household impacts is needed to provide better estimates of the associated costs.
In conclusion, for practitioners and policy-makers concerned about investing in lay health workers for improving the care of patients with non-communicable diseases, our findings demonstrate that the additional investments needed to scale up the MANAS intervention via task-shifting to lay health workers would probably be offset by reduced overall costs. Such an intervention could also provide significant clinical and functional benefits to people with depression and/or anxiety who attend public primary-care facilities. There may be a compelling economic case for investing in lay health workers for the care of other chronic and non-communicable diseases in India. Future studies should be conducted to assess the cost–effectiveness of such an intervention in other settings.
We thank the managements of the Directorate of Health Services, Government of Goa; the Sangath Centre and the Voluntary Health Association of Goa, and the staff of the 24 health facilities.
The MANAS Project is entirely funded by a Wellcome Trust Senior Research Fellowship in Clinical Science awarded to VP.
- Beaglehole R, Epping-Jordan J, Patel V, Chopra M, Ebrahim S, Kidd M, et al., et al. Improving the prevention and management of chronic disease in low-income and middle-income countries: a priority for primary health care. Lancet 2008; 372: 940-9 doi: 10.1016/S0140-6736(08)61404-X pmid: 18790317.
- World Health Organization [Internet]. Task shifting to tackle health worker shortages. Geneva: World Health Organization; 2007. Available from: www.who.int/healthsystems/task_shifting_booklet.pdf [accessed 24 August 2012].
- Mdege ND, Chindove S, Ali S. The effectiveness and cost implications of task-shifting in the delivery of antiretroviral therapy to HIV-infected patients: a systematic review. Health Policy Plan 2012 doi: 10.1093/heapol/czs058.
- Labhardt ND, Balo J-R, Ndam M, Manga E, Stoll B. Improved retention rates with low-cost interventions in hypertension and diabetes management in a rural African environment of nurse-led care: a cluster-randomised trial. Trop Med Int Health 2011; 16: 1276-84 doi: 10.1111/j.1365-3156.2011.02827.x.
- Rahman A, Malik A, Sikander S, Roberts C, Creed F. Cognitive behaviour therapy-based intervention by community health workers for mothers with depression and their infants in rural Pakistan: a cluster-randomised controlled trial. Lancet 2008; 372: 902-9 doi: 10.1016/S0140-6736(08)61400-2 pmid: 18790313.
- Petersen I, Ssebunnya J, Bhana A, Baillie K. Lessons from case studies of integrating mental health into primary health care in South Africa and Uganda. Int J Ment Health Syst 2011; 5: 8- doi: 10.1186/1752-4458-5-8 pmid: 21496242.
- World Health Organization [Internet]. WHO/Wonca joint report: integrating mental health into primary care – a global perspective. Geneva: World Health Organization; 2008. Available from: http://www.who.int/mental_health/policy/services/mentalhealthintoprimarycare/en/ [accessed 2 November 2011].
- Bolton P, Bass J, Neugebauer R, Verdeli H, Clougherty KF, Wickramaratne P, et al., et al. Group interpersonal psychotherapy for depression in rural Uganda: a randomized controlled trial. JAMA 2003; 289: 3117-24 doi: 10.1001/jama.289.23.3117 pmid: 12813117.
- Bass J, Neugebauer R, Clougherty KF, Verdeli H, Wickramaratne P, Ndogoni L, et al., et al. Group interpersonal psychotherapy for depression in rural Uganda: 6-month outcomes: randomised controlled trial. Br J Psychiatry 2006; 188: 567-73 doi: 10.1192/bjp.188.6.567 pmid: 16738348.
- Fulton BD, Scheffler RM, Sparkes SP, Auh EY, Vujicic M, Soucat A. Health workforce skill mix and task shifting in low income countries: a review of recent evidence. Hum Resour Health 2011; 9: 1- doi: 10.1186/1478-4491-9-1 pmid: 21223546.
- Goldberg D, Huxley P. Common mental disorders: a biosocial model. London: Routledge; 1992.
- Mathers C, Boerma T, Fat DM. Global burden of disease: 2004 update. Geneva: World Health Organization; 2008.
- The ICD-10 classification of mental and behavioural disorders. Geneva: World Health Organization; 1992.
- Jacob KS, Everitt BS, Patel V, Weich S, Araya R, Lewis GH. The comparison of latent variable models of non-psychotic psychiatric morbidity in four culturally diverse populations. Psychol Med 1998; 28: 145-52 doi: 10.1017/S0033291797005710 pmid: 9483690.
- Lewis G. Dimensions of neurosis. Psychol Med 1992; 22: 1011-8 doi: 10.1017/S0033291700038575 pmid: 1488473.
- Tyrer P. The case for cothymia: mixed anxiety and depression as a single diagnosis. Br J Psychiatry 2001; 179: 191-3 doi: 10.1192/bjp.179.3.191 pmid: 11532793.
- Araya R, Flynn T, Rojas G, Fritsch R, Simon G. Cost-effectiveness of a primary care treatment program for depression in low-income women in Santiago, Chile. Am J Psychiatry 2006; 163: 1379-87 doi: 10.1176/appi.ajp.163.8.1379 pmid: 16877650.
- Patel V, Weiss HA, Chowdhary N, Naik S, Pednekar S, Chatterjee S, et al., et al. Effectiveness of an intervention led by lay health counsellors for depressive and anxiety disorders in primary care in Goa, India (MANAS): a cluster randomised controlled trial. Lancet 2010; 376: 2086-95 doi: 10.1016/S0140-6736(10)61508-5 pmid: 21159375.
- Patel V, Weiss HA, Chowdhary N, Naik S, Pednekar S, Chatterjee S, et al., et al. Lay health worker led intervention for depressive and anxiety disorders in India: impact on clinical and disability outcomes over 12 months. Br J Psychiatry 2011; 199: 459-66 doi: 10.1192/bjp.bp.111.092155 pmid: 22130747.
- Pereira B, Andrew G, Pednekar S, Kirkwood BR, Patel V. The integration of the treatment for common mental disorders in primary care: experiences of health care providers in the MANAS trial in Goa, India. Int J Ment Health Syst. 2011; 5: 26- doi: 10.1186/1752-4458-5-26 pmid: 21968202.
- Patel VH, Kirkwood BR, Pednekar S, Araya R, King M, Chisholm D, et al., et al. Improving the outcomes of primary care attenders with common mental disorders in developing countries: a cluster randomized controlled trial of a collaborative stepped care intervention in Goa, India. Trials 2008; 9: 4- doi: 10.1186/1745-6215-9-4 pmid: 18221516.
- World Health Organization [Internet]. Cost-effectiveness thresholds. Geneva: World Health Organization; 2011. Available from: http://www.who.int.ezproxy.welch.jhmi.edu/choice/costs/CER_thresholds/en/ [accessed 2 November 2011].http://www.who.int.ezproxy.welch.jhmi.edu/choice/costs/CER_thresholds/en/
- Patel V, Araya R, Chowdhary N, King M, Kirkwood B, Nayak S, et al., et al. Detecting common mental disorders in primary care in India: a comparison of five screening questionnaires. Psychol Med 2008; 38: 221-8 doi: 10.1017/S0033291707002334 pmid: 18047768.
- Lewis G, Pelosi AJ, Araya R, Dunn G. Measuring psychiatric disorder in the community: a standardized assessment for use by lay interviewers. Psychol Med 1992; 22: 465-86 doi: 10.1017/S0033291700030415 pmid: 1615114.
- Patel V, Chisholm D, Rabe-Hesketh S, Dias-Saxena F, Andrew G, Mann A. Efficacy and cost-effectiveness of drug and psychological treatments for common mental disorders in general health care in Goa, India: a randomised, controlled trial. Lancet 2003; 361: 33-9 doi: 10.1016/S0140-6736(03)12119-8 pmid: 12517464.
- Patel V, Pereira J, Coutinho L, Fernandes R, Fernandes J, Mann A. Poverty, psychological disorder and disability in primary care attenders in Goa, India. Br J Psychiatry 1998; 172: 533-6 doi: 10.1192/bjp.172.6.533 pmid: 9828996.
- Patel V, Kirkwood BR, Pednekar S, Pereira B, Barros P, Fernandes J, et al., et al. Gender disadvantage and reproductive health risk factors for common mental disorders in women: a community survey in India. Arch Gen Psychiatry 2006; 63: 404-13 doi: 10.1001/archpsyc.63.4.404 pmid: 16585469.
- Lewis G, Pelosi AJ, Glover E, Wilkinson G, Stansfeld SA, Williams P, et al., et al. The development of a computerized assessment for minor psychiatric disorder. Psychol Med 1988; 18: 737-45 doi: 10.1017/S0033291700008448 pmid: 3054992.
- Üstün TB, Chatterji S, Kostanjsek N, Rehm J, Kennedy C, Epping-Jordan J, et al., et al. Developing the World Health Organization Disability Assessment Schedule 2.0. Bull World Health Organ 2010; 88: 815-23 doi: 10.2471/BLT.09.067231 pmid: 21076562.
- World Health Organization [Internet]. WHO multi-country survey study on health and responsiveness. Geneva: World Health Organization; 2001. Available from: www.who.int/healthinfo/survey/whspaper37.pdf [accessed 2 November 2011].
- Chisholm D, Sekar K, Kumar KK, Saeed K, James S, Mubbashar M, et al., et al. Integration of mental health care into primary care: demonstration cost-outcome study in India and Pakistan. Br J Psychiatry 2000; 176: 581-8 doi: 10.1192/bjp.176.6.581 pmid: 10974966.
- Patel V, Chisholm D, Kirkwood BR, Mabey D. Prioritizing health problems in women in developing countries: comparing the financial burden of reproductive tract infections, anaemia and depressive disorders in a community survey in India. Trop Med Int Health 2007; 12: 130-9 pmid: 17207157.
- Drummond M, Sculpher M, Torrance GW, O’Brien B, Stoddart G. Methods for the economic evaluation of health care programmes. Oxford: Oxford University Press; 2005.
- Mihaylova B, Briggs A, O’Hagan A, Thompson SG. Review of statistical methods for analysing healthcare resources and costs. Health Econ 2011; 20: 897-916 doi: 10.1002/hec.1653 pmid: 20799344.
- Manning WG, Basu A, Mullahy J. Generalized modeling approaches to risk adjustment of skewed outcomes data. J Health Econ 2005; 24: 465-88 doi: 10.1016/j.jhealeco.2004.09.011 pmid: 15811539.
- Glick H, Doshi J, Sonnad S, Polsky D. Economic evaluation in clinical trials. Oxford: Oxford University Press; 2007.
- Hayes R, Moulton L. Cluster randomized trials: a practical approach. Boca Raton: CRC Press; 2009.
- Gomes M, Grieve R, Nixon R, Edmunds WJ. Statistical methods for cost-effectiveness analyses that use data from cluster randomized trials: a systematic review and checklist for critical appraisal. Med Decis Making 2012; 32: 209-20 doi: 10.1177/0272989X11407341 pmid: 21610256.
- Little R, Rubin D. Statistical analysis with missing data. New York: John Wiley & Sons; 1987.
- Baltussen RMPM, Hutubessy RCW, Evans DB, Murray CJM. Uncertainty in cost-effectiveness analysis: probabilistic uncertainty analysis and stochastic league tables. Int J Technol Assess Health Care 2002; 18: 112-9 pmid: 11987434.
- Fenwick E, Marshall D, Levy A, Nichol G. Using and interpreting cost-effectiveness acceptability curves: an example using data from a trial of management strategies for atrial fibrillation. BMC Health Serv Res 2006; 6: 52- doi: 10.1186/1472-6963-6-52 pmid: 16623946.
- Briggs AH. A Bayesian approach to stochastic cost-effectiveness analysis. An illustration and application to blood pressure control in type 2 diabetes. Int J Technol Assess Health Care 2001; 17: 69-82 doi: 10.1017/S0266462301104071 pmid: 11329846.
- Walker D, Fox-Rushby J. Allowing for uncertainty in economic evaluations: qualitative sensitivity analysis. Health Policy Plan 2001; 16: 435-43 doi: 10.1093/heapol/16.4.435 pmid: 11739369.
- Shillcutt SD, Walker DG, Goodman CA, Mills AJ. Cost effectiveness in low- and middle-income countries: a review of the debates surrounding decision rules. Pharmacoeconomics 2009; 27: 903-17 doi: 10.2165/10899580-000000000-00000 pmid: 19888791.
- Government of India. Wage cell. New Delhi: Ministry of Labour and Employment. Available from: http://labour.nic.in/wagecell/welcome.html [accessed 2 November 2011].
- Kakuma R, Minas H, van Ginneken N, Dal Poz MR, Desiraju K, Morris JE, et al., et al. Human resources for mental health care: current situation and strategies for action. Lancet 2011; 378: 1654-63 doi: 10.1016/S0140-6736(11)61093-3 pmid: 22008420.
- Babigumira JB, Castelnuovo B, Stergachis A, Kiragga A, Shaefer P, Lamorde M, et al., et al. Cost effectiveness of a pharmacy-only refill program in a large urban HIV/AIDS clinic in Uganda. PLoS ONE 2011; 6: e18193- doi: 10.1371/journal.pone.0018193 pmid: 21464895.
- Bhandari A, Wagner T. Self-reported utilization of health care services: improving measurement and accuracy. Med Care Res Rev 2006; 63: 217-35 doi: 10.1177/1077558705285298 pmid: 16595412.
- Chisholm D, Knapp MRJ, Knudsen HC, Amaddeo F, Gaite L, Wijngaarden BV. Client Socio-Demographic and Service Receipt Inventory – European Version : development of an instrument for international research EPSILON Study 5. Br J Psychiatry Suppl 2000; 177: s28-33 doi: 10.1192/bjp.177.39.s28 pmid: 10945075.