Safety and Efficacy of Fludrocortisone in the Treatment of Cerebral Salt Wasting in Patients With Tuberculous Meningitis: A Randomized Clinical Trial

Key Points

Question  Is fludrocortisone better than saline in treating cerebral salt wasting with tuberculous meningitis?

Findings  In this randomized clinical trial, 36 patients with tuberculous meningitis with cerebral salt wasting were randomized to receive either an 0.9% solution of intravenous saline and 5 to 12 grams of oral salt supplementation per day or saline, oral salt, and 0.1 to 0.4 mg fludrocortisone per day. Patients treated with fludrocortisone normalized serum sodium levels in 4 days, which was significantly earlier than those receiving saline only (15 days).

Meaning  Fludrocortisone may result in earlier normalization of serum sodium in patients with cerebral salt wasting as a part of tuberculous meningitis.

Importance  Tuberculous meningitis is associated with high frequency of cerebral salt wasting. There is a paucity of objective information regarding the best method of treatment of this condition.

Objective  To evaluate the efficacy and safety of fludrocortisone in the treatment of cerebral salt wasting in patients with tuberculous meningitis.

Design, Setting, and Participants  This is a single-center, open-label, randomized clinical trial conducted from October 2015 to April 2017 in India. Patients were randomized in a 1:1 ratio to arms receiving saline only or saline plus fludrocortisone, in addition to a standard treatment of 4 antitubercular drugs, prednisolone, and aspirin. The 2 arms were matched for demographic, clinical, and magnetic resonance imaging findings. The patients were followed up for at least 6 months.

Interventions  Patients were randomized to a 0.9% solution of intravenous saline with 5 to 12 g per day of oral salt supplementation, with or without the addition of 0.1 to 0.4 mg of fludrocortisone per day.

Main Outcomes and Measures  The primary end point was the time needed to correct serum sodium levels; secondary end points were in-hospital deaths, disability at 3 months, disability at 6 months, occurence of stroke, and serious adverse reactions.

Results  Ninety-three patients with suspected tuberculous meningitis were recruited; 12 did not meet the inclusion criteria, including 4 with alternate diagnoses. A total of 37 patients with cerebral salt wasting were eligible for the study. One refused to participate, and therefore 36 patients were included, with 18 randomized to each group. The median (range) age was 30 (20-46) years, and 19 were male (52.8%). Those receiving fludrocortisone regained normal serum sodium levels after 4 days, significantly earlier than those receiving saline only (15 days; P = .004). In an intention-to-treat analysis, hospital mortality, disability at 3 months, and disability at 6 months did not differ significantly, but fewer infarcts occurred in the deep border zone in the group receiving fludrocortisone (1 of 18 [6%]) vs those in the control arm (6 of 18 [33%]; P = .04). Fludrocortisone was associated with severe hypokalemia and hypertension in 2 patients each, and pulmonary edema occurred in 1 patient. These adverse reactions necessitated discontinuation of fludrocortisone in 2 patients.

Conclusions and Relevance  Fludrocortisone results in earlier normalization of serum sodium levels, but did not affect outcomes at 6 months. Fludrocortisone had to be withdrawn in 2 patients because of severe adverse effects. This study provides class II evidence on the role of fludrocortisone in treatment of hyponatremia associated with cerebral salt wasting in patients with tuberculous meningitis.

Trial Registration  Clinical Trials Registry of India (ctri.nic.in) Identifier: CTRI/2017/10/010255

Cerebral salt wasting (CSW) is defined as the renal loss of sodium because of intracranial diseases leading to hyponatremia, excessive natriuresis, and volume depletion that responds to volume and salt replacement.¹ Most patients with hyponatremia with normal renal function were initially attributed to the syndrome of inappropriate secretion of antidiuretic hormone (SIADH).^2-5 Recently, CSW has been reported to be more common than SIADH in patients experiencing stroke,⁶ acute encephalitis syndrome,⁷ tuberculous meningitis (TBM),⁸ head injury,⁹ and aneurysmal subarachnoid hemorrhage (SAH).¹⁰ In a study on hyponatremia in 76 patients with TBM, CSW occurred in 17 patients (22%); other causes of hyponatremia were SIADH in 3 (4%) and miscellaneous complications in 14 patients (18%), including diuretics in 6 patients, vomiting in 4 patients, and poor intake and endocrine abnormalities in 2 patients each.⁸ In SAH, delayed cerebral ischemia has been attributed to volume contraction because of excessive natriuresis.^11-19 Stroke in patients with TBM may be attributed to volume contraction as a result of CSW, at least in some cases, especially in strokes occurring in the internal border zone.²⁰

In CSW, there is inhibition of renin angiotensin aldosterone system; hence, fludrocortisone has been recommended for treatment of refractory CSW.^21-23 There is limited experience on the role of fludrocortisone in the treatment of CSW, which is mainly based on isolated case reports.^24-28 The only randomized clinical trial on the role of fludrocortisone is in the patients with SAH, which revealed the benefit of fludrocortisone on sodium balance and reduction in delayed stroke.¹⁰ There is no randomized clinical trial evaluating the role of fludrocortisone in patients with CSW associated with TBM. In the present study, we report the efficacy and safety of fludrocortisone in patients with CSW associated with TBM.

This is an investigator-initiated, open-label, randomized clinical trial conducted during October 2015 to April 2017 at a tertiary care teaching hospital in Lucknow, Uttar Pradesh, India. The project was approved by the local ethics committee at Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow, India, on September 30, 2015, and patient enrollment began on October 3, 2015. The patients or their authorized representatives gave written consent. The trial was registered at the Clinical Trials Registry of India (CTRI) on July 1, 2016, 9 months after the trial began due to technical and administrative problems with the CTRI website, and at that time, the Indian Council of Medical Research permitted retrospective trial registration. Seven of 36 patients had been enrolled prior to trial registration; however, no interim analysis was conducted prior to trial registration. The trial protocol is available in Supplement 1.

The patients with TBM who had CSW according to the predefined criteria were included. The diagnosis of TBM was based on clinical, magnetic resonance imaging (MRI), and cerebrospinal fluid (CSF) criteria.²⁹

Essential criteria were features suggestive of meningitis (1 or more of the following: headache, irritability, vomiting, fever, weight loss, neck stiffness, convulsions, focal neurological deficits, or altered consciousness) for more than 5 days. Supportive criteria included (1) 10 to 500 CSF cells per microliter, with predominant lymphocytes (>50%), 1 g of protein per liter, and sterile bacterial and fungal culture; (2) a cranial computed tomographic scan or MRI showing evidence of exudates, infarction, hydrocephalus, or tuberculoma, in isolation or in combination; (3) evidence of tuberculosis outside the central nervous system (eg, a chest radiograph suggestive of active tuberculosis or computed tomographic, MRI, or ultrasonographic evidence of tuberculosis outside the central nervous system, or acid-fast bacilli identified or Mycobacterium tuberculosis cultured from another source, such as sputum, lymph node, gastric washing, or urine), and (4) exclusion of alternative diagnoses.

Findings of essential criteria with 2 supportive criteria were taken to indicate a highly probable diagnosis of TBM. Presence of acid-fast bacilli on CSF smear, a positive CSF culture, or a polymerase chain reaction result positive for M tuberculosis was considered consistent with a definite diagnosis of TBM.²⁹

Cerebral salt wasting was considered in the presence of (1) polyuria (defined as urine output of more than 3 L per day for at least 2 consecutive days); (2) hyponatremia (defined as a serum sodium level less than 135 mEq/L on 2 consecutive evaluations 24 hours apart); and (3) exclusion of secondary causes, such as endocrine abnormalities, use of diuretics, renal failure, cardiac failure, or hepatic failure. All 3 criteria were considered essential to a diagnosis of CSW.

In addition, patients were diagnosed with CSW only if they had at least 3 of the 5 following signs: (1) clinical findings of hypovolemia, such as hypotension, dry mucous membranes, tachycardia, or postural hypotension; (2) persistent negative fluid balance, as determined by intake output record and/or weight loss; (3) laboratory evidence of dehydration, such as elevated hematocrit, hemoglobin, serum albumin, or blood urea nitrogen; (4) a central venous pressure level less than 6 cm of water; and (5) a urinary sodium level greater than 40 mEq/L or urine osmolality greater than 300 mOsm/L on 2 consecutive occasions 24 hours apart.

The patients younger than 10 years; pregnant or lactating women; those with malaria, septic, fungal, or carcinomatous meningitis; and those with head injury, brain tumor, primary failure of the renal, hepatic, or cardiac systems, endocrinal disorders, cancer, or any condition limiting the life expectancy to 1 year or less were excluded. However, patients who developed hepatic or renal dysfunction during treatment were not excluded.

Medical history, including demographic details and information on diabetes and immunosuppression by disease or drugs, was recorded. The duration of illness and presenting symptoms, as well as presence or absence of seizures, focal weakness, evidence of tuberculosis outside the central nervous system, and features of raised intracranial pressure were also noted. Consciousness was assessed using Glasgow Coma Scale. The severity of TBM was categorized into stage I (meningitis only), stage II (meningitis with focal neurological deficit or Glasgow Coma Scale score of 11 to 14), and stage III (meningitis with Glasgow Coma Scale score of less than 11).³⁰

Assessments of each patient included hemoglobin level, liver function tests, serum creatinine level, fasting and postprandial blood glucose levels, HIV serology, radiography of the chest, and abdominal ultrasonography. Cerebrospinal fluid was examined for proteins, cells, glucose level, bacteria, fungi, malignant cells, acid-fast bacilli, and myobacterial culture (BACTEC; Becton Dickinson).

Serum osmolality, urine osmolality, and urine sodium were measured. Serum sodium levels were checked on alternate days until the correction of serum sodium or until the patient was discharged, whichever came earlier. The lowest serum sodium level was used for defining the severity of hyponatremia. Extracellular fluid volume status was assessed on the basis of tachycardia, dry mucous membrane, edema, tenting of the skin, and capillary refill time. Central venous pressure was also measured if indicated. The central venous pressure of 6 to 10 cm of water was considered normal.

A daily fluid intake and output record was maintained. Body weight was measured on admission and monitored daily using a special intensive care bed (Eleganza3XC; LINET). Cranial computed tomographic scan ( Somatom Definition AS Plus; Siemens) and/or MRI (HDxt 32-channel, version 16.0; Signa GE Medical System) were carried out in all the patients on admission. The occurrence of new acute stroke and its location on repeated MRI was also noted. The diagnosis of infarction was based on MRI evidence, which were defined as hyperintensity on T2–weighted and fluid-attenuated inversion recovery MRI, with diffusion restriction on diffusion-weighted imaging.

The patients received antitubercular treatment with 4 drugs (acronymized RHZE, for rifampicin [R], isoniazid [H], pyrazinamide [Z], and ethambutol [E]) for 6 months, followed by a regimen of rifampicin and isoniazid for 12 months. Rifampicin was prescribed at a dose of 10mg/kg (approximately 450 mg per day), isoniazid at a dose of 5 mg/kg (approximately 300 mg per day), pyrazinamide at a dose of 25 mg/kg (approximately 1500 mg per day), and ethambutol at a dose of 15 mg/kg (approximately 800 mg per day). The patients also received prednisone 0.5 mg/kg (approximately 40 mg per day) and aspirin up to 150 mg per day orally, if not contraindicated. The patients were also treated with ventriculoperitoneal shunts or external ventricular drainage or mechanical ventilation if indicated.

All patients with TBM without preexisting comorbidities consistent with exclusion criteria were prescreened for the presence of hyponatremia. Those with hyponatremia underwent detailed clinical and laboratory evaluations, including assessments of weight loss, daily total fluid balance, and urine output, along with urinary sodium and osmolality, serum osmolality, and endocrine function. These patients were categorized as having either CSW (based on predefined criteria) or another cause (extra renal loss, SIADH, or an endocrine disorder). The patients with CSW were randomized to either the control arm (and thus treated with intravenous saline, 0.9%, with oral salt supplementation) or the active arm (and thus given fludrocortisone) in a 1:1 ratio using a computer-generated random number sequence (Figure 1). All patients received intravenous saline, 0.9%, with oral salt supplementation at a dose of 5 to 12 g per day through a nasogastric tube or capsules. The patients in the fludrocortisone arm received fludrocortisone tablets at an oral dose of 0.1 to 0.4 mg once daily, which was started in a dose of 0.1 mg once daily in the morning and was increased every 3 days (from 0.1 mg on day 1 to 0.2 mg on day 4, 0.3 mg on day 7, and 0.4 mg on day 10) until the primary end point was achieved or a total dose of 0.4 mg once daily was reached. Fludrocortisone was discontinued when 2 consecutive serum sodium values (on alternate days) were normal. However, for the primary outcome analysis, the earliest date of serum sodium correction was considered. Failure of the serum sodium level to normalize even after the 0.4-mg dose for 4 days was considered treatment failure; in these cases, fludrocortisone was withdrawn, but intravenous saline with oral salt supplementation was continued. Hypertonic saline in a 3% concentration was used if hyponatremia resulted in seizures or coma.

The primary end point was the number of days necessary for correction of the serum sodium level (≥135 mEq/L). Secondary end points were the number of patients achieving a positive fluid balance, new stroke and its location, time to achieve normal urinary output (<3 L per day), in-hospital mortality, disability at 3 months, and disability at 6 months. The disability was assessed using modified Rankin Scale score and categorized as good (with a score of ≤2 points) or poor (with a score of >2 points). The adverse events of treatment such as hypertension, hypokalemia, and edema were noted.

The patients were followed up at 1, 3, and 6 months for clinical status, activities of daily living, and serum chemistry. A 24-hour urine output was measured once a week, which was also reviewed on the follow-up visits.

Because of the paucity of previous studies, the sample size calculation was based on the assumption of an SD of 2.5 days in the mean days to hyponatremia correction. To detect a difference of 25% between the study and the control arm, with a power of 80% at a significance level of 5% using a 2-sided test, it was calculated that 16 patients in each arm would be needed for sufficient statistical validity.³¹ A dropout rate of 10% was assumed because of the possibility of withdrawal of consent (which would involve patients seeking other treatment options). Thus, a final sample size of 18 patients in each arm was determined.

Continuous and normally distributed variables were represented as mean (SD), whereas the continuous but skewed variables were represented as median and range. Statistical significance was defined as 2-tailed P< .05. For normally distributed continuous variables, we used independent t tests, and for skewed variables, the Mann Whitney U test was used. Fisher exact tests or χ² tests were used to compare categorical variables. Outcomes (including in-hospital deaths, deaths, and disability at 3 and 6 months) were analyzed using an intention-to-treat analysis, which included all 36 patients who consented and were randomly assigned to either the treatment arm or control arm. A prespecified per-protocol analysis was also performed; the population included in the per-protocol analysis was only the patients in both groups who had complied with the treatment protocol without major protocol violations and completed the 6-month follow-up visit. Hierarchical linear regression analysis was used to derive the best predictors (covariates) for the time to hyponatremia correction. All variables with P ≤ .10 for the change in R² were used as covariates in the Cox regression model. Time to hyponatremia correction was modeled using the Cox proportional hazards regression analysis after adjusting the covariates. Statistical analyses were performed using SPSS version 20.0 (IBM).

Ninety-three patients with TBM were prescreened, 37 of whom developed CSW. One patient refused to participate. The results therefore are based on 36 patients (Figure 1). There were 18 patients each in the fludrocortisone and control arms, and their demographic characteristics, baseline clinical findings, and laboratory parameters were similar (Table 1). The median (range) age of the patients was 30 (20-46) years, and 19 were male (52.8%). The median (range) duration of illness was 61 (24-165) days. All the patients received RHZE, aspirin, and prednisolone.

Patients who received fludrocortisone attained normalized serum sodium levels significantly earlier than those who received only saline (4 vs 15 days, respectively; unadjusted P = .004; Figure 2). Serum sodium normalized within 14 days in 15 of the 18 patients (83%) in the fludrocortisone arm, compared with 7 of the 18 patients in the control arm (39%; P = .006; Table 2; Figure 3). The median duration of fludrocortisone treatment was 7 days (range, 4-14 days). One patient in the control arm received saline at a 3% concentration for the correction of severe hyponatremia (109 mEq/L) with convulsions.

A hierarchical linear regression analysis included total oral salt (grams per day), TBM stage, presence of hydrocephalus, sex, use of ventriculoperitoneal shunt, duration of illness in days, use of mechanical ventilation, severity of hyponatremia at the time of randomization, and age to decide the best predictors for the time to correction of hyponatremia. The stage of TBM, severity of hyponatremia at randomization, and age were considered for covariate adjustment (P for change in R² = .08 for TBM stage, .10 for severity of hyponatremia, and .04 for age; eTable 1 in Supplement 2). Time to correction of hyponatremia was longer in the control arm, with a hazard ratio of 4.18 (95% CI, 1.58-11.11; P = .004) on Cox proportional hazards regression analysis after adjusting age, stage of TBM, and severity of hyponatremia (Figure 2).

In intention-to-treat analysis, in-hospital mortality and mortality at 3 months were not significantly different between the 2 arms. Disability at 3 months and disability at 6 months was also similar between the 2 groups (Table 2; eFigures 1, 2, 3, and 4 in Supplement 2). The results of per-protocol analysis were similar to those of the intention-to-treat analysis (eTable 2 in Supplement 2). Magnetic resonance imaging was repeated in 13 patients because of new focal deficits in 12 patients and worsening in sensorium in 1 patient, after a median of 10 (interquartile range, 8-82) days. Five patients developed stroke after fludrocortisone administration (28%), an insignificant increase compared with the control arm (8 patients [44%]; P = .30). The number of infarctions in the internal border zone were significantly lower in the patients receiving fludrocortisone compared with those in the control arm (1 patient [6%] vs 6 patients [33%]; P = .04). All the infarcts occurred during the polyuric phase (Table 2).

The incidence of hypokalemia (< 3.5 mEq/L) was similar in both arms. However, severe hypokalemia (< 2.5 mEq/L) occurred in 2 patients in the fludrocortisone arm, requiring dose reduction in 1 patient and drug withdrawal in the other. Hypokalemia occurred on day 4 (2.2 mEq/L) and day 7 (2.4 mEq/L), which resolved in 8 and 14 days, respectively. Pulmonary edema occurred in 1 patient on day 7 of fludrocortisone treatment, which responded to an intravenous diuretic. Hypertension occurred in 2 patients and responded to the withdrawal of fludrocortisone, and neither patient required antihypertensive drugs (Table 3). The frequency of recurrent hyponatremia after the primary end point was insignificantly lower in patients receiving fludrocortisone compared with those in the control arm (4 patients [22%] vs 7 patients [39%]; P = .28). These patients were treated with saline and oral salt.

In the present study, oral fludrocortisone resulted in earlier normalization of serum sodium compared with saline alone, but it neither ameliorated polyuria nor affected mortality or the development of disability. There was no difference in the frequency of stroke, but there were fewer deep border zone infarcts in the fludrocortisone arm. These results are in agreement with a randomized clinical trial in aneurysmal SAH.¹⁰

In TBM, administration of daily doses of 0.2 to 0.6 mg of fludrocortisone has been attempted in the treatment of CSW in 6 studies^28,32-36 of patients with stage II or III TBM who were aged 6 months to 70 years. Overall, serum sodium levels improved in 0.5 to 10 days. Urinary output also improved in all but 1 patient, whose urine output was not reported.

The treatment of CSW has 2 components: volume replacement and the correction of hyponatremia and natriuresis. Generally a 2% to 3% saline solution is used, which increases serum sodium concentration rapidly at a risk of pulmonary edema, heart failure, and osmotic demyelination. Moreover, the benefit of hypertonic saline is transient, and the pathological stimulus producing natriuresis and hypovolemia persists. In the study of Hasan et al,¹⁰ both negative sodium balance and polyuria responded within 2 weeks and were associated with lower frequency of delayed cerebral ischemia. In this study, although serum sodium levels normalized within 4 (IQR, 3-11) days, polyuria and negative fluid balance persisted for a much longer period (range, 30-320 days) and was not significantly reduced by fludrocortisone. Moreover, an insignificantly smaller number of patients developed recurrence of hyponatremia in the fludrocortisone arm than in the control arm (4 of 18 patients [22%] vs 7 of 18 patients [39%]). The possible reasons for poor response of polyuria might be the relatively short-term use of fludrocortisone. Fludrocortisone was stopped when serum sodium levels normalized. Longer treatment with fludrocortisone could have reduced polyuria and restored the negative fluid balance and recurrence of hyponatremia. In addition, the severe and more protracted stress in TBM compared with SAH might account for the poor response in cases of polyuria. The patients in this study had more severe disease, as evidenced by the presence of tuberculoma in 28 patients, hydrocephalus in 27, and infarctions in 17. The stimulus for CSW therefore persisted and could be refractory to treatment. It has been shown that lowering intracranial pressure by CSF drainage or shunt may be an effective treatment of CSW in addition to fluid and sodium chloride administration.³⁷

The mechanism of CSW has been attributed to a reduced sympathetic drive because of intracranial disease that leads to inhibition of renin angiotensin aldosterone system and failure of aldosterone to rise. This, along with release of natriuretic factors, may result in natriuresis and volume contraction.³⁸ Brain natriuretic peptide has been regarded as a more likely cause of CSW.³⁹ It has been suggested that the release of brain natriuretic peptide may be a stress response to the underlying increased intracranial pressure. A linear relationship between CSF level of this peptide and intracranial pressure has been reported in patients with SAH.⁴⁰

In patients with TBM, stroke occurs in 40%. These strokes can be asymptomatic, and some of them may be in the border zone area.²⁰ The frequency of clinical stroke was similar in both arms of this study. Magnetic resonance imaging of these patients revealed fewer infarcts in the internal border zone among those in the fludrocortisone arm. Delayed vasospasm and stroke are well documented in patients with SAH and have been treated by therapy combining induced hypertension, hypervolemia, and hemodilution.

Prolonged use of fludrocortisone may be associated with serious adverse effects. We observed severe hypokalemia in 2 patients, hypertension in 2 patients, and pulmonary edema in 1 patient. We had to withdraw fludrocortisone in 2 patients. Hasan et al¹⁰ also observed pulmonary edema necessitating fludrocortisone withdrawal in 2 patients; however, pulmonary edema also occurred in 2 of their control patients.

This is an open-label pilot study based on a small sample. Well-characterized patients with CSW are difficult to recruit at a single center. Because of the lack of any previous study, the standard deviation for sample size calculation was arbitrarily chosen to be 2.5 days.

In addition, we used normal saline solution with oral salt supplementation instead of a 3% saline solution, unless there was a medical emergency. We have not used an identical-looking placebo for fludrocortisone, but our end points were sufficiently robust and objective that they are therefore unlikely to be influenced by insufficient blinding.

In addition, we only completed MRI scanning in patients with clinically apparent stroke, and not in all the patients at different points. It is therefore impossible to comment on the possible benefit of fludrocortisone in preventing infarcts.

The frequency of recurrent hyponatremia was lower in the fludrocortisone arm compared with the control arm, although this difference did not reach statistical significance. However, further studies are needed to evaluate the role of fludrocortisone doses and treatment duration in the management of polyuria, its possible effects on stroke, and its role in reducing the frequency of recurrent hyponatremia after fludrocortisone treatment.

Fludrocortisone results in earlier normalization of serum sodium levels in patients with TBM who are experiencing CSW. However, in this study, administration of the drug did not affect polyuria or the overall outcome of TBM.

Corresponding Author: Usha K. Misra, DM, Department of Neurology, Sanjay Gandhi Post Graduate Institute of Medical Sciences, Raebareily Road, Lucknow-226014, Uttar Pradesh, India (drukmisra@rediffmail.com, drukmisra@gmail.com).

Accepted for Publication: June 14, 2018.

Published Online: August 13, 2018. doi:10.1001/jamaneurol.2018.2178

Correction: This article was corrected on October 15, 2018, to add information to the first paragraph of the Methods section about an administrative delay in trial registration associated with technical failures at the Clinical Trials Registry of India website.

Author Contributions: Drs Misra and Kumar have full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Misra.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Misra, Kalita.

Critical revision of the manuscript for important intellectual content: Kumar.

Statistical analysis: Kalita, Kumar.

Obtained funding: Misra.

Supervision: Misra, Kalita.

Conflict of Interest Disclosures: None reported.

Funding/support: The project was funded by Indian Council of Medical Research grant 2013-21330.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contribution: We thank all the patients who participated in the study. We also thank S. K. Mandal, MSc, Centre for Biomedical Research (CBMR) for help in statistical analysis and Shakti Kumar, MSc, Sanjay Gandhi Post Graduate Institute of Medical Sciences, for secretarial help. Mr Shakti received a stipend from the Indian Council of Medical Research for his contribution, and Mr Mandal was not compensated for his contribution.